inkling
Thinking Machines/🇺🇸 US/chat
Inkling is a large MoE hybrid reasoning model from Thinking Machines with audio and vision input support and a 64K context window.MoreLess
Which id to call
thinkingmachines/inklingThis exact deployment on Thinking Machines, with no routing and no failover. Send it as the model field.
Input /1M
$1.87
$0.37 cached
Output /1M
$4.68
2.5x input
Context
66K
33K output
Added
Jul 2026
chat
Capabilities 4/8
Provider rates
What inkling costs
Provider prices per 1M tokens, updated October 19, 2026.moreless
These are the upstream provider rates. Pay as you go adds 5%, or 0% if you bring your own keys, and there is no per-request fee. Prompt caching and routing change what you pay against these rates, not the rates themselves.
Input /1M
$1.87
Output /1M
$4.68
Cache write /1M
-
Cache read /1M
$0.37
What a workload costs
- 100K input + 10K output
- $0.23
- 1M input + 100K output
- $2.34
- 10M input + 1M output
- $23.38
At the rates above, before caching. A cache read costs $0.37 per 1M, so repeated context lands under these figures.
model=
Which id to call
One base url, https://router.requesty.ai/v1, and one key for every id here.
thinkingmachines/inklingThis exact deployment on Thinking Machines, with no routing and no failover. Send it as the model field.
OpenAI compatible
Call it in three lines
Change the base url, use your Requesty key, set the model to any id on the left. Existing OpenAI SDK code needs no other edit.
123456789101112131415from openai import OpenAI client = OpenAI( api_key="YOUR_REQUESTY_API_KEY", base_url="https://router.requesty.ai/v1", ) response = client.chat.completions.create( model="thinkingmachines/inkling", messages=[ {"role": "user", "content": "Explain quantum computing in one paragraph."}, ], ) print(response.choices[0].message.content)
Reference
Specs and data terms
What the catalog reports for this deployment, and what the provider does with the traffic.
Released 2026-07-15
Benchmark scores
Benchmarks measure the model, not this endpoint, so they are the same wherever these weights are served.
Artificial Analysis Coding Index: a composite of coding evaluations including LiveCodeBench, SciCode and Terminal-Bench.
Graduate-level physics, chemistry & biology questions designed to resist Googling.
Artificial Analysis Intelligence Index: a composite of multiple evaluations measuring overall model capability.
Scores from official model cards, Artificial Analysis and public leaderboards. They measure specific skills and do not capture every aspect of model quality, so test on your own workload.
Same provider
More from Thinking Machines
Newest first, on the same provider and the same key.
Reference
inkling questions
How much does inkling cost?
inkling is priced at $1.87 per million input tokens and $4.68 per million output tokens when accessed via Requesty. Prompt caching is supported, which can cut effective input cost by up to 90% on repeated context. Those are the upstream provider rates: pay as you go adds 5% on top, or 0% if you bring your own provider keys.
What is the context window of inkling?
inkling has a context window of 66K tokens, with a maximum output of 33K tokens per response. That's roughly 87 words of input you can fit in a single prompt.
How does inkling perform on benchmarks?
inkling scores 87.2% on GPQA Diamond, 52.1% on Coding Index, 47.0% on SciCode. See the full benchmark chart above for results across MMLU Pro, GPQA Diamond, SWE-Bench Verified, HumanEval, MATH, AIME, MMMU, and LiveBench.
What can inkling do?
inkling supports vision input, tool calling, extended reasoning, prompt caching. You can call it through any OpenAI-compatible client by pointing base_url to Requesty.
How do I use inkling with the OpenAI SDK?
Install the OpenAI SDK, set base_url to "https://router.requesty.ai/v1", set your API key to your Requesty key, and set the model to "thinkingmachines/inkling". The Quickstart above shows Python, JavaScript and cURL snippets.
Can I run inkling through Requesty?
Yes. inkling runs through Requesty's OpenAI-compatible API, served from Thinking Machines. You do not host the model yourself: point base_url at Requesty, set the model to "thinkingmachines/inkling", and requests are routed to the upstream provider with automatic failover. The same key gives you 600+ other models too.
Call inkling through one endpoint
One key for this endpoint and 600+ other models. No markup on provider prices, automatic failover, caching built in.