{
  "@context": "https://schema.org",
  "@type": "Dataset",
  "@id": "https://www.requesty.ai/rankings",
  "name": "The State of Production AI: model share, cost and speed",
  "description": "Share of tokens, spend and requests per model version, with blended cost per million tokens, cost per session, cache hit rates and streaming latency, measured from routed production traffic.",
  "version": "1",
  "generatedAt": "2026-08-03T17:37:39.521355982Z",
  "window": {
    "from": "2026-06-01",
    "to": "2026-08-03",
    "weeks": [
      "2026-06-01",
      "2026-06-08",
      "2026-06-15",
      "2026-06-22",
      "2026-06-29",
      "2026-07-06",
      "2026-07-13",
      "2026-07-20",
      "2026-07-27",
      "2026-08-03"
    ]
  },
  "platform": {
    "blended_cost_per_mtok_usd": 1.02,
    "cache_hit_rate_pct": 67.9,
    "cache_savings_pct": 61.7,
    "reasoning_request_pct": 23.3,
    "stream_request_pct": 24.8,
    "model_count": 40,
    "provider_count": 29,
    "withheld_share_pct": 2.9
  },
  "regions": [
    "ap",
    "eu",
    "us"
  ],
  "columns": [
    {
      "key": "rank",
      "label": "rank",
      "note": "Position by token share over the window"
    },
    {
      "key": "model",
      "label": "model",
      "note": "Model version as routed"
    },
    {
      "key": "vendor",
      "label": "vendor",
      "note": "Model vendor"
    },
    {
      "key": "line",
      "label": "product_line",
      "note": "Vendor product line"
    },
    {
      "key": "token_share_pct",
      "label": "token_share_pct",
      "note": "Share of all tokens processed over the window"
    },
    {
      "key": "spend_share_pct",
      "label": "spend_share_pct",
      "note": "Share of all model spend over the window"
    },
    {
      "key": "request_share_pct",
      "label": "request_share_pct",
      "note": "Share of all requests over the window"
    },
    {
      "key": "token_share_delta_pp",
      "label": "token_share_delta_pp",
      "note": "Week over week change in token share, percentage points"
    },
    {
      "key": "blended_cost_per_mtok_usd",
      "label": "blended_cost_per_mtok_usd",
      "note": "Charged price per million tokens, prompt cache discounts included"
    },
    {
      "key": "median_session_cost_usd",
      "label": "median_session_cost_usd",
      "note": "Median spend on one session, trace-id clients only"
    },
    {
      "key": "cost_per_request_usd",
      "label": "cost_per_request_usd",
      "note": "Mean spend on one request"
    },
    {
      "key": "cache_hit_rate_pct",
      "label": "cache_hit_rate_pct",
      "note": "Prompt cache hit rate"
    },
    {
      "key": "cache_savings_pct",
      "label": "cache_savings_pct",
      "note": "Share of the bill removed by caching"
    },
    {
      "key": "median_ttft_ms",
      "label": "median_ttft_ms",
      "note": "Median time to first token, streaming responses only"
    },
    {
      "key": "p95_ttft_ms",
      "label": "p95_ttft_ms",
      "note": "p95 time to first token, streaming responses only"
    },
    {
      "key": "median_tokens_per_sec",
      "label": "median_tokens_per_sec",
      "note": "Median output speed, streaming responses only"
    },
    {
      "key": "tokens_per_request",
      "label": "tokens_per_request",
      "note": "Mean tokens per request"
    },
    {
      "key": "reasoning_request_pct",
      "label": "reasoning_request_pct",
      "note": "Share of requests that returned reasoning tokens"
    },
    {
      "key": "stream_request_pct",
      "label": "stream_request_pct",
      "note": "Share of requests streamed"
    },
    {
      "key": "median_session_requests",
      "label": "calls_per_session",
      "note": "Median calls in one session"
    },
    {
      "key": "median_session_minutes",
      "label": "session_minutes",
      "note": "Median session length in minutes"
    },
    {
      "key": "error_rate_5xx_pct",
      "label": "error_rate_5xx_pct",
      "note": "Server error rate"
    }
  ],
  "models": [
    {
      "rank": 1,
      "model": "claude-sonnet-4-6",
      "vendor": "anthropic",
      "line": "sonnet",
      "token_share_pct": 11.6,
      "spend_share_pct": 18.5,
      "request_share_pct": 5.4,
      "token_share_delta_pp": 2.2,
      "blended_cost_per_mtok_usd": 1.64,
      "median_session_cost_usd": 0.35,
      "cost_per_request_usd": 0.081289,
      "cache_hit_rate_pct": 62.7,
      "cache_savings_pct": 57.3,
      "median_ttft_ms": 1468,
      "p95_ttft_ms": 4179,
      "median_tokens_per_sec": 77,
      "tokens_per_request": 49621,
      "reasoning_request_pct": 0.7,
      "stream_request_pct": 54.8,
      "median_session_requests": 6,
      "median_session_minutes": 4.1,
      "error_rate_5xx_pct": 0.1
    },
    {
      "rank": 2,
      "model": "deepseek-v4-flash",
      "vendor": "deepseek",
      "line": "deepseek",
      "token_share_pct": 11.3,
      "spend_share_pct": 0.4,
      "request_share_pct": 5.5,
      "token_share_delta_pp": 1.4,
      "blended_cost_per_mtok_usd": 0.03,
      "median_session_cost_usd": 0.03,
      "cost_per_request_usd": 0.001507,
      "cache_hit_rate_pct": 87,
      "cache_savings_pct": 77.6,
      "median_ttft_ms": 1200,
      "p95_ttft_ms": 7560,
      "median_tokens_per_sec": 79,
      "tokens_per_request": 47592,
      "reasoning_request_pct": 59.1,
      "stream_request_pct": 48,
      "median_session_requests": 4,
      "median_session_minutes": 0.9,
      "error_rate_5xx_pct": 0.5
    },
    {
      "rank": 3,
      "model": "claude-opus-4-8",
      "vendor": "anthropic",
      "line": "opus",
      "token_share_pct": 9.7,
      "spend_share_pct": 16.6,
      "request_share_pct": 2.2,
      "token_share_delta_pp": 2,
      "blended_cost_per_mtok_usd": 1.75,
      "median_session_cost_usd": 1.51,
      "cost_per_request_usd": 0.175132,
      "cache_hit_rate_pct": 81.6,
      "cache_savings_pct": 76.4,
      "median_ttft_ms": 1739,
      "p95_ttft_ms": 5826,
      "median_tokens_per_sec": 87,
      "tokens_per_request": 100255,
      "reasoning_request_pct": 2.6,
      "stream_request_pct": 70.5,
      "median_session_requests": 12,
      "median_session_minutes": 6.7,
      "error_rate_5xx_pct": 0.2
    },
    {
      "rank": 4,
      "model": "deepseek-v4-pro",
      "vendor": "deepseek",
      "line": "deepseek",
      "token_share_pct": 6.8,
      "spend_share_pct": 0.8,
      "request_share_pct": 2.5,
      "token_share_delta_pp": -3.8,
      "blended_cost_per_mtok_usd": 0.12,
      "median_session_cost_usd": 0.04,
      "cost_per_request_usd": 0.00749,
      "cache_hit_rate_pct": 91.3,
      "cache_savings_pct": 79.6,
      "median_ttft_ms": 1000,
      "p95_ttft_ms": 3828,
      "median_tokens_per_sec": 56,
      "tokens_per_request": 61578,
      "reasoning_request_pct": 75.5,
      "stream_request_pct": 62.2,
      "median_session_requests": 5,
      "median_session_minutes": 19.5,
      "error_rate_5xx_pct": 0.4
    },
    {
      "rank": 5,
      "model": "glm-5.2",
      "vendor": "zai",
      "line": "glm",
      "token_share_pct": 6.2,
      "spend_share_pct": 3.2,
      "request_share_pct": 1.9,
      "token_share_delta_pp": 0,
      "blended_cost_per_mtok_usd": 0.53,
      "median_session_cost_usd": 0.39,
      "cost_per_request_usd": 0.039754,
      "cache_hit_rate_pct": 81.9,
      "cache_savings_pct": 62,
      "median_ttft_ms": 1936,
      "p95_ttft_ms": 16513,
      "median_tokens_per_sec": 105,
      "tokens_per_request": 74464,
      "reasoning_request_pct": 35.5,
      "stream_request_pct": 83.2,
      "median_session_requests": 13,
      "median_session_minutes": 3.5,
      "error_rate_5xx_pct": 1.1
    },
    {
      "rank": 6,
      "model": "claude-sonnet-5",
      "vendor": "anthropic",
      "line": "sonnet",
      "token_share_pct": 5.5,
      "spend_share_pct": 4.4,
      "request_share_pct": 1.5,
      "token_share_delta_pp": 3.4,
      "blended_cost_per_mtok_usd": 0.83,
      "median_session_cost_usd": 0.58,
      "cost_per_request_usd": 0.070305,
      "cache_hit_rate_pct": 76.9,
      "cache_savings_pct": 71.4,
      "median_ttft_ms": 1363,
      "p95_ttft_ms": 3975,
      "median_tokens_per_sec": 99,
      "tokens_per_request": 85085,
      "reasoning_request_pct": 7.9,
      "stream_request_pct": 60.4,
      "median_session_requests": 10,
      "median_session_minutes": 3.1,
      "error_rate_5xx_pct": 0.1
    },
    {
      "rank": 7,
      "model": "gemini-3.5-flash",
      "vendor": "google",
      "line": "gemini-flash",
      "token_share_pct": 4.6,
      "spend_share_pct": 4.2,
      "request_share_pct": 3.1,
      "token_share_delta_pp": 0,
      "blended_cost_per_mtok_usd": 0.95,
      "median_session_cost_usd": 0.06,
      "cost_per_request_usd": 0.032283,
      "cache_hit_rate_pct": 57.5,
      "cache_savings_pct": 48.4,
      "median_ttft_ms": 2030,
      "p95_ttft_ms": 13751,
      "median_tokens_per_sec": 489,
      "tokens_per_request": 34084,
      "reasoning_request_pct": 54,
      "stream_request_pct": 56.3,
      "median_session_requests": 4,
      "median_session_minutes": 0.3,
      "error_rate_5xx_pct": 0.1
    },
    {
      "rank": 8,
      "model": "gemini-2.5-flash",
      "vendor": "google",
      "line": "gemini-flash",
      "token_share_pct": 3.8,
      "spend_share_pct": 3,
      "request_share_pct": 9.1,
      "token_share_delta_pp": 0.3,
      "blended_cost_per_mtok_usd": 0.81,
      "median_session_cost_usd": null,
      "cost_per_request_usd": 0.007763,
      "cache_hit_rate_pct": 4.8,
      "cache_savings_pct": 1.2,
      "median_ttft_ms": 1573,
      "p95_ttft_ms": 9418,
      "median_tokens_per_sec": 3347,
      "tokens_per_request": 9549,
      "reasoning_request_pct": 71.5,
      "stream_request_pct": 1,
      "median_session_requests": 3,
      "median_session_minutes": null
    },
    {
      "rank": 9,
      "model": "gpt-5.4",
      "vendor": "openai",
      "line": "gpt-5",
      "token_share_pct": 3.5,
      "spend_share_pct": 6.6,
      "request_share_pct": 3.3,
      "token_share_delta_pp": 0.5,
      "blended_cost_per_mtok_usd": 1.9,
      "median_session_cost_usd": 0.27,
      "cost_per_request_usd": 0.047443,
      "cache_hit_rate_pct": 53.5,
      "cache_savings_pct": 38,
      "median_ttft_ms": 989,
      "p95_ttft_ms": 57096,
      "median_tokens_per_sec": 102,
      "tokens_per_request": 24970,
      "reasoning_request_pct": 9.5,
      "stream_request_pct": 56.6,
      "median_session_requests": 7,
      "median_session_minutes": 0.7,
      "error_rate_5xx_pct": 0.2
    },
    {
      "rank": 10,
      "model": "claude-haiku-4-5",
      "vendor": "anthropic",
      "line": "haiku",
      "token_share_pct": 3.3,
      "spend_share_pct": 4.1,
      "request_share_pct": 8,
      "token_share_delta_pp": -0.5,
      "blended_cost_per_mtok_usd": 1.26,
      "median_session_cost_usd": 0.03,
      "cost_per_request_usd": 0.012001,
      "cache_hit_rate_pct": 36,
      "cache_savings_pct": 20.7,
      "median_ttft_ms": 661,
      "p95_ttft_ms": 1823,
      "median_tokens_per_sec": 128,
      "tokens_per_request": 9507,
      "reasoning_request_pct": 0.1,
      "stream_request_pct": 14.6,
      "median_session_requests": 4,
      "median_session_minutes": 0.3,
      "error_rate_5xx_pct": 0.2
    },
    {
      "rank": 11,
      "model": "gpt-5.5",
      "vendor": "openai",
      "line": "gpt-5",
      "token_share_pct": 3.2,
      "spend_share_pct": 7.3,
      "request_share_pct": 1.4,
      "token_share_delta_pp": 0.3,
      "blended_cost_per_mtok_usd": 2.32,
      "median_session_cost_usd": 0.73,
      "cost_per_request_usd": 0.126336,
      "cache_hit_rate_pct": 77.6,
      "cache_savings_pct": 62,
      "median_ttft_ms": 11289,
      "p95_ttft_ms": 58684,
      "median_tokens_per_sec": 215,
      "tokens_per_request": 54485,
      "reasoning_request_pct": 68.8,
      "stream_request_pct": 55.3,
      "median_session_requests": 9,
      "median_session_minutes": 4.7,
      "error_rate_5xx_pct": 0.4
    },
    {
      "rank": 12,
      "model": "gemini-2.5-flash-lite",
      "vendor": "google",
      "line": "gemini-flash-lite",
      "token_share_pct": 2.3,
      "spend_share_pct": 0.3,
      "request_share_pct": 11,
      "token_share_delta_pp": -0.1,
      "blended_cost_per_mtok_usd": 0.13,
      "median_session_cost_usd": null,
      "cost_per_request_usd": 0.000641,
      "cache_hit_rate_pct": 2.1,
      "cache_savings_pct": 1.3,
      "median_ttft_ms": 274,
      "p95_ttft_ms": 1230,
      "median_tokens_per_sec": 263,
      "tokens_per_request": 4894,
      "reasoning_request_pct": 3.2,
      "stream_request_pct": 0.8,
      "median_session_requests": null,
      "median_session_minutes": null,
      "error_rate_5xx_pct": 0.1
    },
    {
      "rank": 13,
      "model": "claude-opus-4-7",
      "vendor": "anthropic",
      "line": "opus",
      "token_share_pct": 2.3,
      "spend_share_pct": 4.7,
      "request_share_pct": 0.6,
      "token_share_delta_pp": -2,
      "blended_cost_per_mtok_usd": 2.11,
      "median_session_cost_usd": 1.98,
      "cost_per_request_usd": 0.200699,
      "cache_hit_rate_pct": 73.9,
      "cache_savings_pct": 66.8,
      "median_ttft_ms": 2060,
      "p95_ttft_ms": 7309,
      "median_tokens_per_sec": 108,
      "tokens_per_request": 94966,
      "reasoning_request_pct": 1.4,
      "stream_request_pct": 72.1,
      "median_session_requests": 13,
      "median_session_minutes": 8.6,
      "error_rate_5xx_pct": 0.2
    },
    {
      "rank": 14,
      "model": "deepseek-chat",
      "vendor": "deepseek",
      "line": "deepseek",
      "token_share_pct": 2,
      "spend_share_pct": 0.2,
      "request_share_pct": 7.4,
      "token_share_delta_pp": -0.1,
      "blended_cost_per_mtok_usd": 0.11,
      "median_session_cost_usd": null,
      "cost_per_request_usd": 0.000679,
      "cache_hit_rate_pct": 40.7,
      "cache_savings_pct": 27.3,
      "median_ttft_ms": 568,
      "p95_ttft_ms": 1351,
      "median_tokens_per_sec": 75,
      "tokens_per_request": 6125,
      "stream_request_pct": 0.5,
      "median_session_requests": null,
      "median_session_minutes": null,
      "error_rate_5xx_pct": 0.1
    },
    {
      "rank": 15,
      "model": "gemini-3.1-pro-preview",
      "vendor": "google",
      "line": "gemini-pro",
      "token_share_pct": 1.9,
      "spend_share_pct": 4.8,
      "request_share_pct": 2.1,
      "token_share_delta_pp": -1,
      "blended_cost_per_mtok_usd": 2.62,
      "median_session_cost_usd": null,
      "cost_per_request_usd": 0.053455,
      "cache_hit_rate_pct": 18.1,
      "cache_savings_pct": 11,
      "median_ttft_ms": 9886,
      "p95_ttft_ms": 50527,
      "median_tokens_per_sec": 118,
      "tokens_per_request": 20437,
      "reasoning_request_pct": 85.9,
      "stream_request_pct": 14,
      "median_session_requests": null,
      "median_session_minutes": null,
      "error_rate_5xx_pct": 3.8
    },
    {
      "rank": 16,
      "model": "gpt-5.6-luna",
      "vendor": "openai",
      "line": "gpt-5",
      "token_share_pct": 1.8,
      "spend_share_pct": 0.6,
      "request_share_pct": 1,
      "token_share_delta_pp": 0.1,
      "blended_cost_per_mtok_usd": 0.36,
      "median_session_cost_usd": 0.15,
      "cost_per_request_usd": 0.015541,
      "cache_hit_rate_pct": 76.2,
      "cache_savings_pct": 63.3,
      "median_ttft_ms": 1677,
      "p95_ttft_ms": 6515,
      "median_tokens_per_sec": 77,
      "tokens_per_request": 42601,
      "reasoning_request_pct": 69.6,
      "stream_request_pct": 47.9,
      "median_session_requests": 16,
      "median_session_minutes": 2.4,
      "error_rate_5xx_pct": 0.4
    },
    {
      "rank": 17,
      "model": "claude-opus-4-6",
      "vendor": "anthropic",
      "line": "opus",
      "token_share_pct": 1.7,
      "spend_share_pct": 5.2,
      "request_share_pct": 0.8,
      "token_share_delta_pp": -0.3,
      "blended_cost_per_mtok_usd": 3.02,
      "median_session_cost_usd": 0.78,
      "cost_per_request_usd": 0.149222,
      "cache_hit_rate_pct": 65.7,
      "cache_savings_pct": 62,
      "median_ttft_ms": 2409,
      "p95_ttft_ms": 6349,
      "median_tokens_per_sec": 67,
      "tokens_per_request": 49355,
      "reasoning_request_pct": 0.7,
      "stream_request_pct": 61.9,
      "median_session_requests": 10,
      "median_session_minutes": 13.8,
      "error_rate_5xx_pct": 0.2
    },
    {
      "rank": 18,
      "model": "minimax-m3",
      "vendor": "minimaxi",
      "line": "minimax",
      "token_share_pct": 1.7,
      "spend_share_pct": 0.2,
      "request_share_pct": 0.8,
      "token_share_delta_pp": -1.3,
      "blended_cost_per_mtok_usd": 0.12,
      "median_session_cost_usd": 0.01,
      "cost_per_request_usd": 0.005997,
      "cache_hit_rate_pct": 91.9,
      "cache_savings_pct": 66.7,
      "median_ttft_ms": 1848,
      "p95_ttft_ms": 7580,
      "median_tokens_per_sec": 90,
      "tokens_per_request": 49052,
      "reasoning_request_pct": 46.9,
      "stream_request_pct": 43.4,
      "median_session_requests": 3,
      "median_session_minutes": 0.3,
      "error_rate_5xx_pct": 0.3
    },
    {
      "rank": 19,
      "model": "gpt-5.6-sol",
      "vendor": "openai",
      "line": "gpt-5",
      "token_share_pct": 1.5,
      "spend_share_pct": 1.9,
      "request_share_pct": 0.3,
      "token_share_delta_pp": -2.5,
      "blended_cost_per_mtok_usd": 1.26,
      "median_session_cost_usd": 1.02,
      "cost_per_request_usd": 0.146155,
      "cache_hit_rate_pct": 87.2,
      "cache_savings_pct": 75.8,
      "median_ttft_ms": 1290,
      "p95_ttft_ms": 6836,
      "median_tokens_per_sec": 46,
      "tokens_per_request": 116184,
      "reasoning_request_pct": 68,
      "stream_request_pct": 84.8,
      "median_session_requests": 9,
      "median_session_minutes": 4.4,
      "error_rate_5xx_pct": 0.3
    },
    {
      "rank": 20,
      "model": "claude-opus-5",
      "vendor": "anthropic",
      "line": "opus",
      "token_share_pct": 1.2,
      "spend_share_pct": 1.7,
      "request_share_pct": 0.2,
      "token_share_delta_pp": 0.4,
      "blended_cost_per_mtok_usd": 1.43,
      "median_session_cost_usd": 2.47,
      "cost_per_request_usd": 0.168123,
      "cache_hit_rate_pct": 88.9,
      "cache_savings_pct": 80.6,
      "median_ttft_ms": 1837,
      "p95_ttft_ms": 4601,
      "median_tokens_per_sec": 78,
      "tokens_per_request": 117943,
      "reasoning_request_pct": 33.4,
      "stream_request_pct": 74,
      "median_session_requests": 21,
      "median_session_minutes": 9,
      "error_rate_5xx_pct": 0.6
    },
    {
      "rank": 21,
      "model": "gemini-3.1-flash-lite",
      "vendor": "google",
      "line": "gemini-flash-lite",
      "token_share_pct": 1.2,
      "spend_share_pct": 0.3,
      "request_share_pct": 5.1,
      "token_share_delta_pp": -0.5,
      "blended_cost_per_mtok_usd": 0.26,
      "median_session_cost_usd": 0.01,
      "cost_per_request_usd": 0.001457,
      "cache_hit_rate_pct": 23.4,
      "cache_savings_pct": 18,
      "median_ttft_ms": 943,
      "p95_ttft_ms": 3633,
      "median_tokens_per_sec": 283,
      "tokens_per_request": 5505,
      "reasoning_request_pct": 1,
      "stream_request_pct": 3.7,
      "median_session_requests": 5,
      "median_session_minutes": 2.7
    },
    {
      "rank": 22,
      "model": "gpt-5.6-terra",
      "vendor": "openai",
      "line": "gpt-5",
      "token_share_pct": 1.2,
      "spend_share_pct": 0.9,
      "request_share_pct": 0.4,
      "token_share_delta_pp": 0.5,
      "blended_cost_per_mtok_usd": 0.82,
      "median_session_cost_usd": 0.2,
      "cost_per_request_usd": 0.055343,
      "cache_hit_rate_pct": 81.9,
      "cache_savings_pct": 68.9,
      "median_ttft_ms": 1254,
      "p95_ttft_ms": 5371,
      "median_tokens_per_sec": 84,
      "tokens_per_request": 67126,
      "reasoning_request_pct": 65.6,
      "stream_request_pct": 75.1,
      "median_session_requests": 5,
      "median_session_minutes": 0.9,
      "error_rate_5xx_pct": 1.7
    },
    {
      "rank": 23,
      "model": "kimi-k3",
      "vendor": "moonshot",
      "line": "kimi",
      "token_share_pct": 1.1,
      "spend_share_pct": 0.8,
      "request_share_pct": 0.4,
      "token_share_delta_pp": 0,
      "blended_cost_per_mtok_usd": 0.75,
      "median_session_cost_usd": 0.63,
      "cost_per_request_usd": 0.05061,
      "cache_hit_rate_pct": 90.9,
      "cache_savings_pct": 74.7,
      "median_ttft_ms": 3524,
      "p95_ttft_ms": 12147,
      "median_tokens_per_sec": 47,
      "tokens_per_request": 67607,
      "reasoning_request_pct": 67.1,
      "stream_request_pct": 64.3,
      "median_session_requests": 15,
      "median_session_minutes": 6.8,
      "error_rate_5xx_pct": 0.3
    },
    {
      "rank": 24,
      "model": "gpt-4o-mini",
      "vendor": "openai",
      "line": "gpt-mini",
      "token_share_pct": 0.9,
      "spend_share_pct": 0.1,
      "request_share_pct": 9.1,
      "token_share_delta_pp": 0,
      "blended_cost_per_mtok_usd": 0.13,
      "median_session_cost_usd": null,
      "cost_per_request_usd": 0.00031,
      "cache_hit_rate_pct": 55.2,
      "cache_savings_pct": 23.8,
      "median_ttft_ms": 1145,
      "p95_ttft_ms": 4686,
      "median_tokens_per_sec": 90,
      "tokens_per_request": 2406,
      "stream_request_pct": 0.4,
      "median_session_requests": null,
      "median_session_minutes": null,
      "error_rate_5xx_pct": 0.6
    },
    {
      "rank": 25,
      "model": "claude-fable-5",
      "vendor": "anthropic",
      "line": "fable",
      "token_share_pct": 0.9,
      "spend_share_pct": 3.5,
      "request_share_pct": 0.2,
      "token_share_delta_pp": 0.3,
      "blended_cost_per_mtok_usd": 4.06,
      "median_session_cost_usd": 2.68,
      "cost_per_request_usd": 0.353596,
      "cache_hit_rate_pct": 74.9,
      "cache_savings_pct": 67.7,
      "median_ttft_ms": 4041,
      "p95_ttft_ms": 8841,
      "median_tokens_per_sec": 80,
      "tokens_per_request": 87062,
      "reasoning_request_pct": 5.9,
      "stream_request_pct": 65.2,
      "median_session_requests": 12,
      "median_session_minutes": 7.5,
      "error_rate_5xx_pct": 0.2
    },
    {
      "rank": 26,
      "model": "kimi-k2.7-code",
      "vendor": "moonshot",
      "line": "kimi",
      "token_share_pct": 0.8,
      "spend_share_pct": 0.2,
      "request_share_pct": 0.3,
      "token_share_delta_pp": 0.7,
      "blended_cost_per_mtok_usd": 0.3,
      "median_session_cost_usd": 0.23,
      "cost_per_request_usd": 0.016767,
      "cache_hit_rate_pct": 94.2,
      "cache_savings_pct": 70.7,
      "median_ttft_ms": 1741,
      "p95_ttft_ms": 6093,
      "median_tokens_per_sec": 92,
      "tokens_per_request": 55478,
      "reasoning_request_pct": 47.2,
      "stream_request_pct": 77.9,
      "median_session_requests": 16,
      "median_session_minutes": 2.7,
      "error_rate_5xx_pct": 0.4
    },
    {
      "rank": 27,
      "model": "gemini-3-flash-preview",
      "vendor": "google",
      "line": "gemini-flash",
      "token_share_pct": 0.6,
      "spend_share_pct": 0.4,
      "request_share_pct": 1.4,
      "token_share_delta_pp": -0.1,
      "blended_cost_per_mtok_usd": 0.64,
      "median_session_cost_usd": 0.11,
      "cost_per_request_usd": 0.006478,
      "cache_hit_rate_pct": 31.6,
      "cache_savings_pct": 20.1,
      "median_ttft_ms": 4601,
      "p95_ttft_ms": 18513,
      "median_tokens_per_sec": 933,
      "tokens_per_request": 10103,
      "reasoning_request_pct": 61.1,
      "stream_request_pct": 12.6,
      "median_session_requests": 9,
      "median_session_minutes": 8.9,
      "error_rate_5xx_pct": 0.1
    },
    {
      "rank": 28,
      "model": "claude-sonnet-4-5",
      "vendor": "anthropic",
      "line": "sonnet",
      "token_share_pct": 0.6,
      "spend_share_pct": 1.1,
      "request_share_pct": 0.7,
      "token_share_delta_pp": 0.1,
      "blended_cost_per_mtok_usd": 2.03,
      "median_session_cost_usd": 0.31,
      "cost_per_request_usd": 0.037046,
      "cache_hit_rate_pct": 53.6,
      "cache_savings_pct": 49.2,
      "median_ttft_ms": 1711,
      "p95_ttft_ms": 4186,
      "median_tokens_per_sec": 64,
      "tokens_per_request": 18229,
      "reasoning_request_pct": 0.4,
      "stream_request_pct": 25,
      "median_session_requests": 8,
      "median_session_minutes": 1.8,
      "error_rate_5xx_pct": 0.2
    },
    {
      "rank": 29,
      "model": "gpt-5.4-mini",
      "vendor": "openai",
      "line": "gpt-mini",
      "token_share_pct": 0.5,
      "spend_share_pct": 0.3,
      "request_share_pct": 1.4,
      "token_share_delta_pp": 0,
      "blended_cost_per_mtok_usd": 0.63,
      "median_session_cost_usd": null,
      "cost_per_request_usd": 0.004912,
      "cache_hit_rate_pct": 37.7,
      "cache_savings_pct": 28.1,
      "median_ttft_ms": 692,
      "p95_ttft_ms": 1869,
      "median_tokens_per_sec": 141,
      "tokens_per_request": 7798,
      "reasoning_request_pct": 3.9,
      "stream_request_pct": 48,
      "median_session_requests": 2,
      "median_session_minutes": null,
      "error_rate_5xx_pct": 0.1
    },
    {
      "rank": 30,
      "model": "gpt-5.1",
      "vendor": "openai",
      "line": "gpt-5",
      "token_share_pct": 0.4,
      "spend_share_pct": 0.5,
      "request_share_pct": 1,
      "token_share_delta_pp": 0,
      "blended_cost_per_mtok_usd": 1.2,
      "median_session_cost_usd": 0.03,
      "cost_per_request_usd": 0.012215,
      "cache_hit_rate_pct": 22.8,
      "cache_savings_pct": 17.3,
      "median_ttft_ms": 533,
      "p95_ttft_ms": 3264,
      "median_tokens_per_sec": 118,
      "tokens_per_request": 10162,
      "reasoning_request_pct": 14.2,
      "stream_request_pct": 59.7,
      "median_session_requests": 4,
      "median_session_minutes": 0.8,
      "error_rate_5xx_pct": 0.3
    },
    {
      "rank": 31,
      "model": "minimax-m2.5",
      "vendor": "minimaxi",
      "line": "minimax",
      "token_share_pct": 0.4,
      "spend_share_pct": 0.1,
      "request_share_pct": 0.3,
      "token_share_delta_pp": 0.1,
      "blended_cost_per_mtok_usd": 0.12,
      "median_session_cost_usd": 0.04,
      "cost_per_request_usd": 0.003738,
      "cache_hit_rate_pct": 75.5,
      "cache_savings_pct": 60.6,
      "median_ttft_ms": 795,
      "p95_ttft_ms": 4083,
      "median_tokens_per_sec": 76,
      "tokens_per_request": 31467,
      "reasoning_request_pct": 0.1,
      "stream_request_pct": 75.6,
      "median_session_requests": 28.5,
      "median_session_minutes": 5.3,
      "error_rate_5xx_pct": 0.9
    },
    {
      "rank": 32,
      "model": "gpt-4.1-mini",
      "vendor": "openai",
      "line": "gpt-mini",
      "token_share_pct": 0.3,
      "spend_share_pct": 0.1,
      "request_share_pct": 2.5,
      "token_share_delta_pp": -0.3,
      "blended_cost_per_mtok_usd": 0.31,
      "median_session_cost_usd": null,
      "cost_per_request_usd": 0.001024,
      "cache_hit_rate_pct": 30.5,
      "cache_savings_pct": 28.3,
      "median_ttft_ms": 505,
      "p95_ttft_ms": 2045,
      "median_tokens_per_sec": 84,
      "tokens_per_request": 3279,
      "stream_request_pct": 1.9,
      "median_session_requests": null,
      "median_session_minutes": null,
      "error_rate_5xx_pct": 1.4
    },
    {
      "rank": 33,
      "model": "glm-5.1",
      "vendor": "zai",
      "line": "glm",
      "token_share_pct": 0.3,
      "spend_share_pct": 0.3,
      "request_share_pct": 0.2,
      "token_share_delta_pp": 0,
      "blended_cost_per_mtok_usd": 0.82,
      "median_session_cost_usd": 0.18,
      "cost_per_request_usd": 0.035904,
      "cache_hit_rate_pct": 53.5,
      "cache_savings_pct": 41.9,
      "median_ttft_ms": 2070,
      "p95_ttft_ms": 12915,
      "median_tokens_per_sec": 44,
      "tokens_per_request": 44017,
      "reasoning_request_pct": 9,
      "stream_request_pct": 86.7,
      "median_session_requests": 9,
      "median_session_minutes": 2,
      "error_rate_5xx_pct": 6.2
    },
    {
      "rank": 34,
      "model": "kimi-k2.6",
      "vendor": "moonshot",
      "line": "kimi",
      "token_share_pct": 0.3,
      "spend_share_pct": 0.1,
      "request_share_pct": 0.2,
      "token_share_delta_pp": -0.1,
      "blended_cost_per_mtok_usd": 0.44,
      "median_session_cost_usd": 0.25,
      "cost_per_request_usd": 0.017674,
      "cache_hit_rate_pct": 74.4,
      "cache_savings_pct": 53.5,
      "median_ttft_ms": 1652,
      "p95_ttft_ms": 8629,
      "median_tokens_per_sec": 48,
      "tokens_per_request": 40432,
      "reasoning_request_pct": 14.2,
      "stream_request_pct": 84.2,
      "median_session_requests": 10,
      "median_session_minutes": 6,
      "error_rate_5xx_pct": 2.6
    },
    {
      "rank": 35,
      "model": "grok-4.5",
      "vendor": "xai",
      "line": "grok",
      "token_share_pct": 0.3,
      "spend_share_pct": 0.3,
      "request_share_pct": 0.1,
      "token_share_delta_pp": -0.5,
      "blended_cost_per_mtok_usd": 0.83,
      "median_session_cost_usd": 0.22,
      "cost_per_request_usd": 0.069957,
      "cache_hit_rate_pct": 94.3,
      "cache_savings_pct": 70.5,
      "median_ttft_ms": 1487,
      "p95_ttft_ms": 5197,
      "median_tokens_per_sec": 75,
      "tokens_per_request": 84163,
      "reasoning_request_pct": 89.3,
      "stream_request_pct": 58.2,
      "median_session_requests": 7.5,
      "median_session_minutes": 5.7,
      "error_rate_5xx_pct": 3
    },
    {
      "rank": 36,
      "model": "hy3",
      "vendor": "tencent",
      "line": "hunyuan",
      "token_share_pct": 0.3,
      "request_share_pct": 0.1,
      "token_share_delta_pp": -0.3,
      "blended_cost_per_mtok_usd": 0.03,
      "median_session_cost_usd": null,
      "cost_per_request_usd": 0.00154,
      "cache_hit_rate_pct": 75.6,
      "cache_savings_pct": 57,
      "median_ttft_ms": 4155,
      "p95_ttft_ms": 16029,
      "median_tokens_per_sec": 61,
      "tokens_per_request": 57527,
      "reasoning_request_pct": 60.4,
      "stream_request_pct": 71.7,
      "median_session_requests": 2,
      "median_session_minutes": 23.8,
      "error_rate_5xx_pct": 0.4
    },
    {
      "rank": 37,
      "model": "mimo-v2.5",
      "vendor": "xiaomi",
      "line": "mimo",
      "token_share_pct": 0.3,
      "request_share_pct": 0.1,
      "token_share_delta_pp": -0.1,
      "blended_cost_per_mtok_usd": 0.02,
      "median_session_cost_usd": null,
      "cost_per_request_usd": 0.001691,
      "cache_hit_rate_pct": 90.5,
      "cache_savings_pct": 84.6,
      "median_ttft_ms": 1384,
      "p95_ttft_ms": 5014,
      "median_tokens_per_sec": 82,
      "tokens_per_request": 71476,
      "reasoning_request_pct": 88.2,
      "stream_request_pct": 90.7,
      "median_session_requests": null,
      "median_session_minutes": null,
      "error_rate_5xx_pct": 1
    },
    {
      "rank": 38,
      "model": "nemotron-3-ultra-550b-a55b",
      "vendor": "nvidia",
      "line": "nemotron",
      "token_share_pct": 0.3,
      "request_share_pct": 0.1,
      "token_share_delta_pp": -0.2,
      "blended_cost_per_mtok_usd": null,
      "median_session_cost_usd": null,
      "cost_per_request_usd": 0.000212,
      "cache_hit_rate_pct": 56.8,
      "median_ttft_ms": 2533,
      "p95_ttft_ms": 22143,
      "median_tokens_per_sec": 65,
      "tokens_per_request": 49025,
      "reasoning_request_pct": 3.1,
      "stream_request_pct": 67.2,
      "median_session_requests": 2,
      "median_session_minutes": 1,
      "error_rate_5xx_pct": 9
    },
    {
      "rank": 39,
      "model": "gpt-5-mini",
      "vendor": "openai",
      "line": "gpt-mini",
      "token_share_pct": 0.3,
      "spend_share_pct": 0.1,
      "request_share_pct": 0.8,
      "token_share_delta_pp": 0.1,
      "blended_cost_per_mtok_usd": 0.29,
      "median_session_cost_usd": null,
      "cost_per_request_usd": 0.002151,
      "cache_hit_rate_pct": 59.1,
      "cache_savings_pct": 29.3,
      "median_ttft_ms": 3137,
      "p95_ttft_ms": 13796,
      "median_tokens_per_sec": 106,
      "tokens_per_request": 7320,
      "reasoning_request_pct": 79.6,
      "stream_request_pct": 26.1,
      "median_session_requests": 2,
      "median_session_minutes": 0.4,
      "error_rate_5xx_pct": 0.3
    },
    {
      "rank": 40,
      "model": "deepseek-v4-flash-0731",
      "vendor": "deepseek",
      "line": "deepseek",
      "token_share_pct": 0.2,
      "request_share_pct": 0.2,
      "token_share_delta_pp": 4.6,
      "blended_cost_per_mtok_usd": 0.07,
      "median_session_cost_usd": 0.08,
      "cost_per_request_usd": 0.001713,
      "cache_hit_rate_pct": 86.9,
      "cache_savings_pct": 46.7,
      "median_ttft_ms": 330,
      "p95_ttft_ms": 1401,
      "median_tokens_per_sec": 133,
      "tokens_per_request": 24992,
      "reasoning_request_pct": 0.1,
      "stream_request_pct": 67.8,
      "median_session_requests": 21,
      "median_session_minutes": 4.8,
      "error_rate_5xx_pct": 0.4
    }
  ],
  "methodology": [
    "Every figure is a share or a per-unit rate. No request, token, session or spend totals are published, so nothing here sums back into platform volume or revenue.",
    "Each model version is its own row, because moving from one version to the next is the decision teams make. The product line is reported alongside for grouping.",
    "A model, provider and week is published only when enough separate organizations used it and no single one dominated it. 2.9% of traffic in this window failed those floors and is excluded, which is why shares do not sum to exactly 100.",
    "Time to first token and output speed come from streaming responses only, because a non streaming request has no first token event.",
    "A session is a run of calls sharing one trace id with no gap over 30 minutes. Only clients that send a trace id appear, and the figure is a median because session cost has a long tail.",
    "Cost per million tokens is what was charged with prompt cache discounts included, which is why it sits under provider list prices.",
    "An empty cell is a withheld or unsampled metric, not a zero."
  ],
  "distribution": {
    "json": "https://www.requesty.ai/rankings/data.json",
    "csv": "https://www.requesty.ai/rankings/data.csv",
    "markdown": "https://www.requesty.ai/rankings/data.md"
  },
  "image": "https://www.requesty.ai/rankings/opengraph-image",
  "license": "Free to cite with attribution to Requesty.",
  "source": {
    "organization": "Requesty",
    "url": "https://www.requesty.ai"
  }
}