[whatmodel.sh](../../README.md) / Inkling (975B-A41B) # Inkling (975B-A41B) Thinking Machines · released 2026-07 · 975B params (41B active) · apache-2.0 **score 67.8** ±0.7 · #103 of 732 · from 35 benchmark results across 6 sources Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only. ## What it's good at _{● marks Inkling (975B-A41B) among every model on that task or test, worst to best.}

task                vs the others         rank  counts  tests
agent tasks         ▁▁▃▆█▄▅●▄▄▂▁   #102 of 313     25%     11
coding              ▁▁▁▃▅▇█▇●▄▃▁    #86 of 469     12%      4
chat & writing      ▁▁▁▃▆▄▆█▆▇●▃   #61 of 402*    4.5%      1
math                ▁▁▃▅█▆▄●▄▂▁▁    #92 of 431     33%      5
vision & documents  ▁▁▂▂▁▃▄▂▆●█▄   #70 of 153*    2.7%      2
web search          ●··········█      #2 of 2*    2.2%      2

agent tasks · #102 of 313 · 25% of its score

Beats [Grok 4.20 (0309 reasoning)](grok-4-20.md), [o3-pro](o3-pro.md), [Grok 4.3](grok-4-3.md) here, though they rank higher overall. Trails [Nova 2 Pro (Preview)](nova-2-pro.md), [Mistral Medium 3.5 (128B)](mistral-medium-3-5.md), [Hy3 preview (295B-A21B)](hy3-preview.md) here, though they rank lower overall.

test                                     result  vs the others in its run    difficulty            counts  why
Arena Agent (October 2026 aggrega…  -0.1580 IPS  ▁●▁▃▃▂█▁▂▂▁▁ beats 2/59     avg -0.379 IPS           12%  sharp
SWE-bench Verified (maker)                77.6%  ▁▁▁▁▁▁▂▄█●▁▁ beats 120/138  hard · avg 28%          2.6%  sharp · small run
τ³-bench banking_knowledge (v1.0.…        25.0%  ▃▃▂▃·●▃▂█▅▅▂ beats 8/25     very hard · avg 4.3%    2.3%  typical
Terminal-Bench 2.1 (maker)                63.8%  ▁·▁·▁▁▄▃▆●▅█ beats 30/63    hard · avg 19%          2.2%  sharp · at its level · small run
MCP-Atlas (public set) (maker)            78.8%  ▂·▁▂▂·▁▃█▆●▂ beats 35/39    medium · avg 39%        1.3%  at its level · small run
Humanity's Last Exam with tools (…        46.0%  ▁·▁▂▂▁▄▄●█▄▅ beats 26/62    hard · avg 16%          1.2%  at its level · small run
OSWorld-Verified                          40.1%  ●▃·▂▅▂▃▂▅▃▆▆ beats 4/27     hard · avg 19%          1.1%  at its level · small run
BrowseComp with context managemen…        77.1%  ▄·▄▂·●▂·▆▄·▂ beats 7/15     medium · avg 46%        1.0%  small run
Toolathlon-Verified (maker)               45.5%  ▂·▂▂●▃···▂██ beats 3/18     hard · avg 28%          0.7%  blunt · at its level · small run
SWE-bench Pro (maker)                     54.3%  ▄▂▂▂●▂▆█·▂·▂ beats 5/16     hard · avg 31%          0.6%  blunt · at its level · small run
τ³-bench Banking (maker)                  23.7%  ▂▂·▅█▂▃▃▃●▂▅ beats 20/24    very hard · avg 6.6%    0.5%  blunt · small run

coding · #86 of 469 · 12% of its score

Beats [Grok 4.20 (0309 reasoning)](grok-4-20.md), [GPT-5.4 mini](gpt-5-4-mini.md), [GPT-5.1](gpt-5-1.md) here, though they rank higher overall. Trails [Llama 3.1 Nemotron Ultra 253B v1](llama-3-1-nemotron-ultra-253b.md), [GLM-4.7](glm-4-7.md) here, though they rank lower overall.

test                                result  vs the others in its run    difficulty      counts  why
Arena WebDev (Code Arena)           1415.2  ▁▁▃▂▅●▃▆▅▅▁▂ beats 49/129   avg 981           8.0%  sharp
SWE-bench Verified (maker)           77.6%  ▁▁▁▁▁▁▂▄█●▁▁ beats 120/138  hard · avg 28%    2.6%  sharp · small run
SciCode (level unspecified) (make…   46.1%  ▁····▃▃▄▅█●▂ beats 36/40    hard · avg 18%    1.1%  at its level · small run
SWE-bench Pro (maker)                54.3%  ▄▂▂▂●▂▆█·▂·▂ beats 5/16     hard · avg 31%    0.6%  blunt · at its level · small run

chat & writing · #61 of 402 · 4.5% of its score

Beats [Claude Haiku 5.5](claude-haiku-5-5.md), [GPT-6 Luna](gpt-6-luna.md), [Grok 4.7](grok-4-7.md) here, though they rank higher overall. Trails [Gemma 4 31B](gemma-4-31b.md), [Gemini 2.5 Pro](gemini-2-5-pro.md), [Claude Sonnet 4.5](claude-sonnet-4-5.md) here, though they rank lower overall.

test            result  vs the others in its run    difficulty  counts  why
LMArena (text)  1442.2  ▁▁▁▂▃▃▃▆▆▇●▃ beats 308/385  avg 1260      4.5%  blunt

math · #92 of 431 · 33% of its score

Beats [Claude Sonnet 4.6](claude-sonnet-4-6.md), [Claude Opus 4.5](claude-opus-4-5.md), [MiniMax-M3](minimax-m3.md) here, though they rank higher overall. Trails [Gemini 2.5 Flash](gemini-2-5-flash.md), [Granite 4.2 30B](granite-4-2-30b.md), [Qwen3-Next 80B-A3B Instruct](qwen3-next-80b-a3b-instruct.md) here, though they rank lower overall.

test                        result  vs the others in its run    difficulty            counts  why
FrontierMath tiers 1–3 v2    33.3%  ▂··▁▁▃●▄█▆▄▃ beats 21/86    very hard · avg 1.2%     19%  sharp · at its level
FrontierMath tier 4 v2        4.9%  ▁▂●▂█▅▂▃▂▁▁▁ beats 10/67    very hard · avg 0.1%    6.5%  sharp
HMMT February 2026 (maker)   86.3%  ▃▃▂▅·████●▃▅ beats 28/35    medium · avg 41%        2.6%  sharp · small run
AIME 2026 (maker)            97.1%  ▁▁▁▂▂▂▆█▆●·▁ beats 50/52    easy · avg 69%          2.5%  sharp · maxed out · small run
OTIS mock AIME 2024–2025     88.9%  ▃▃▄▃▃▃▇▇●▃▂▅ beats 140/197  very hard · avg 7.2%    2.1%  sharp · noisy

vision & documents · #70 of 153 · 2.7% of its score

Beats [DeepSeek V4 Flash Vision (Exp)](deepseek-v4-flash-vision-exp.md), [GPT-5](gpt-5.md), [o3](o3.md) here, though they rank higher overall. Trails [Gemma 4 26B-A4B](gemma-4-26b-a4b.md), [Claude Haiku 4.5](claude-haiku-4-5.md), [Qwen3.6 Flash](qwen3-6-flash.md) here, though they rank lower overall.

test                                result  vs the others in its run   difficulty        counts  why
CharXiv (reasoning questions) (ma…   78.1%  ▂▅▃▅▅▆▆▅●█▆▂ beats 73/101  medium · avg 40%    1.5%  small run
MMMU Pro (maker)                     73.5%  ▂▃▁▃▆▇▃▇▇●█▅ beats 35/52   medium · avg 45%    1.2%  at its level · small run

web search · #2 of 2 · 2.2% of its score

test                                result  vs the others in its run  difficulty        counts  why
Humanity's Last Exam with tools (…   46.0%  ▁·▁▂▂▁▄▄●█▄▅ beats 26/62  hard · avg 16%      1.2%  at its level · small run
BrowseComp with context managemen…   77.1%  ▄·▄▂·●▂·▆▄·▂ beats 7/15   medium · avg 46%    1.0%  small run

other tests (15)

test                                result  vs the others in its run    difficulty            counts  why
GPQA Diamond                         88.3%  ▁▁▁▁▂▄▆▄▅▇●▅ beats 162/211  hard · avg 39%          9.3%  sharp · several runs
SimpleQA Verified                    40.3%  ▁▁▃▂▃▃▆●█▄▂▅ beats 33/81    very hard · avg 8.0%    2.5%  at its level · several runs
Chess Puzzles (Epoch)                21.0%  █▁▁▁▁▂●▃▃▂▁▁ beats 85/137   very hard · avg 5.8%    2.3%  sharp · noisy
Humanity's Last Exam (maker)         29.7%  ▁▁▁▄▆▆▇█●▆▁▃ beats 74/111   very hard · avg 6.2%    1.8%  at its level · small run
Icelandic LLM Leaderboard average…   46.5%  ▁▁▃▂▂▄▁▃●▄▄█ beats 17/38    very hard · avg 14%     1.6%  at its level
Global-MMLU-Lite (maker)             88.7%  ▂···▂·▂▂▂·●█ beats 8/12     medium · avg 60%        1.6%  small run
CritPt (maker)                        5.4%  ▄···▄▄▄●█▄▄▄ beats 4/9      very hard · avg 1.3%    1.4%  too hard for it · small run
SimpleVQA with tools (search)        52.3%  ●·········██ beats 0/2      hard · avg 27%          1.2%  at its level · small run
Video-MME v2 (with subtitles)        36.9%  ●·········██ beats 0/2      hard · avg 17%          1.1%  at its level · small run
IFBench (aggregation unspecified)…   79.8%  ▂▆▂▄▄▄▂▄▄▆█● beats 80/86    medium · avg 47%        1.1%  small run
IntPhys 2                            55.2%  ●···█······█ beats 0/2      hard · avg 27%          0.9%  at its level · small run
BabyVision with tools                34.2%  ●······█··██ beats 0/3      hard · avg 30%          0.5%  at its level · noisy · small run
ZeroBench main with tools (pass@5)   14.0%  ●········█·▄ beats 0/3                                 —  not counted yet: no other model in the …
MMAU (maker)                         77.2%  █··········● beats 1/1                                 —  not counted yet: no other model in the …
WorldVQA with tools                  18.0%  ●··········█ beats 0/1                                 —  not counted yet: no other model in the …

## How the score was worked out Each result implies a score on its own, after adjusting for how hard that test is and who ran it. The model's score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its *counts* column above; here it is added up by who ran the tests. [Method](../method.md)

source                                    results  share of the score
Epoch AI                                        6   38% ▇▇▇▇▇
thinkingmachines/Inkling-Small model ca…       19   29% ▇▇▇▇
LMArena leaderboard dataset                     3   24% ▇▇▇
Meta vendor-reported benchmark scores f…        5    5% ▇
τ²/τ³-bench leaderboard: Sierra's own e…        1    2% ▇
Miðeind Icelandic LLM Leaderboard               1    2% ▇
prior: pull toward the middle                        0% ▇
weighted average                                   = 67.8

_{Licences: Epoch AI ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · thinkingmachines/Inkling-Small model card ([Apache-2.0](../../licenses/thinkingmachines--Inkling-Small.txt)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · τ²/τ³-bench leaderboard: Sierra's own evaluation runs ([MIT](https://opensource.org/license/mit)) · Miðeind Icelandic LLM Leaderboard ([CC BY 4.0](../../licenses/mideind-icelandic-llm-leaderboard-CC-BY-4.0.txt))}

Not counted (3)

source                  test                            score  why
thinkingmachines/Inkl…  MMAU                            77.2%  too few models have it yet
Meta vendor-reported …  ZeroBench main with tools (pa…  14.0%  too few models have it yet
Meta vendor-reported …  WorldVQA with tools             18.0%  too few models have it yet

## How to run it ### Through an API api (Deep Infra) · $1.72/M blended · $0.95 in / $4.05 out per 1M tokens · [deepinfra.com](https://deepinfra.com/models) _{[price source](https://github.com/anomalyco/models.dev/blob/a668908cfe7073faa0363699486fc41a3ff52412/providers/deepinfra/models/thinkingmachines/Inkling.toml): Standard base token rate from models.dev. Provider-specific tiers, long-context surcharges, caching and promotions may differ. · [context source](https://github.com/anomalyco/models.dev/blob/a668908cfe7073faa0363699486fc41a3ff52412/models/thinkingmachines/inkling.toml)} ### llama.cpp (recommended) 1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md) ```sh # macOS / Linux curl -LsSf https://llama.app/install.sh | sh # Windows irm https://llama.app/install.ps1 | iex ``` 2. Run Inkling (975B-A41B) (ud-q4_k_xl, needs ~585.5 GB, from [huggingface.co/unsloth](https://huggingface.co/unsloth/inkling-GGUF/tree/main)) ```sh llama cli -hf unsloth/inkling-GGUF:UD-Q4_K_XL ``` 3. Or start a local server and open the address it prints (http://localhost:9931) ```sh llama serve -hf unsloth/inkling-GGUF:UD-Q4_K_XL ``` Other versions: ud-q2_k_xl (~293 GB) `llama cli -hf unsloth/inkling-GGUF:UD-Q2_K_XL` · ud-iq1_s (~146.75 GB) `llama cli -hf unsloth/inkling-GGUF:UD-IQ1_S` ### LM Studio 1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download) 2. Run Inkling (975B-A41B) (ud-q4_k_xl, needs ~585.5 GB, from [huggingface.co/unsloth](https://huggingface.co/unsloth/inkling-GGUF/tree/main)) ```sh lms get https://huggingface.co/unsloth/inkling-GGUF@UD-Q4_K_XL -y ``` 3. Then run lms ls to see its name, and lms chat followed by that name Other versions: ud-q2_k_xl (~293 GB) `lms get https://huggingface.co/unsloth/inkling-GGUF@UD-Q2_K_XL -y` · ud-iq1_s (~146.75 GB) `lms get https://huggingface.co/unsloth/inkling-GGUF@UD-IQ1_S -y` Model repository and documentation: [huggingface.co/thinkingmachines](https://huggingface.co/thinkingmachines/Inkling) [interactive version](https://whatmodel.sh/model/inkling) · [JSON](https://whatmodel.sh/api?format=json&all=1)