Thinking Machines 发布 Inkling 2 Small，权重将开源为评分最高的中国以外开放权重模型

Thinking Machines has released Inkling 2 Small, scoring 31 on the Artificial Analysis Intelligence Index - their plan to release its weights would make it the highest-scoring open weights model developed outside China based on current rankings

See model page

Inkling 2 Small is launching as a proprietary model, available first-party through Thinking Machines' Tinker platform and Tinker Playground, with weights to be released at a later date. Its score is up 18 points from Inkling Small, with ~4x as many output tokens per task. It matches Google DeepMind's Gemini 3.8 Flash (high) and surpasses Xiaomi's MiMo-V2.6-Pro on the Artificial Analysis Intelligence Index v5.0.1.

Key takeaways:

➤ Large gains over Inkling Small, especially on agentic tasks: AutomationBench-AA rises from 5% to 46% and FrontierCode 2.0 from 5% to 25%. HLE-Diamond also rises from 12% to 29%.

➤ Different coding strengths from Claude Sonnet 5.5 (medium): Inkling 2 Small scores 25% on FrontierCode 2.0 vs. 11% for Claude Sonnet 5.5 (medium), but 13% on Terminal-Bench 4.0 vs. 30%.

➤ Creative agentic scores above Claude Haiku 5.5 and Grok 4.7 on AA-3D: Inkling 2 Small also scores above Gemini 3.8 Flash on both AA-3D and Code2Video-AA, but behind Claude Sonnet 5.5 on both.

➤ Research-level physics improves, but trails Gemini 3.8 Flash (high): Inkling 2 Small scores 56% on CritPt v2, up from 34% for Inkling Small, vs. 65% for Gemini 3.8 Flash (high).

➤ Fewer hallucinations from more selective answering: On AA-Omniscience v1.1, the hallucination rate falls from Inkling Small's 61% to 27%, while accuracy rises from 32% to 34% and the attempt rate falls from 77% to 53%.

➤ ~4x as many output tokens: Inkling 2 Small averages ~150K output tokens per Intelligence Index task vs. ~37K for Inkling Small, ~31K for Claude Sonnet 5.5 (medium), and ~99K for Gemini 3.8 Flash (high).

➤ Almost 6x the cost of Inkling Small, but cheaper than several other models with similar Intelligence Index scores: Inkling 2 Small costs $1.18 per Intelligence Index task vs. $0.20 for Inkling Small, $7.77 for Gemini 3.8 Flash (high), and $10.02 for Claude Sonnet 5.5 (medium).

Key model details:

➤ Type: Proprietary reasoning model at launch; Apache 2.0 weights release planned (date unconfirmed)

➤ Modalities: Text and image input, text output

➤ Size: 594B total parameters / 17B active parameters

➤ Pricing: $0.30/$1.00 per 1M input/output tokens ($0.03 per 1M cached input tokens)

➤ Availability at launch: Tinker and Tinker Playground

Congratulations to the team at Thinking Machines on the release!

Inkling 2 Small scores 615 Elo points higher than Inkling Small on AA-Briefcase, with 2.5x the rubric pass rate.

AA-Briefcase tests long-horizon business workflows requiring spreadsheets, presentations, and memos. Inkling 2 Small scores 1527 Elo vs. 912 Elo for Inkling Small, 1439 Elo for Claude Sonnet 5.5 (medium), and 1203 Elo for Gemini 3.8 Flash (high).

It passes 50% of rubric checks vs. 20% for Inkling Small, with analytical quality at 1634 Elo and presentation at 1564 Elo.

Inkling 2 Small makes large gains over Inkling (xhigh) on AA-3D and Code2Video-AA, with scores close to Muse Spark 1.3 and Qwen3.8 Max, respectively.

AA-3D tests models’ ability to create interactive 3D experiences. Inkling 2 Small scores 1021 Elo vs. 966 Elo for Google DeepMind's Gemini 3.8 Flash and 1033 Elo for Meta's Muse Spark 1.3 (max).

On Code2Video-AA, which tests models’ ability to create videos using code, Inkling 2 Small scores 1020 Elo vs. 899 Elo for Gemini 3.8 Flash and 1021 Elo for Alibaba's Qwen3.8 Max (0902) (xhigh).

Compared with Inkling (xhigh), it gains 438 Elo on AA-3D (1021 vs. 583) and 553 Elo on Code2Video-AA (1020 vs. 467).

Inkling 2 Small hallucinates less mainly because it answers fewer questions, while accuracy improves only slightly.

On AA-Omniscience v1.1 its net score improves to 16 from -9 for Inkling Small, but remains behind Gemini 3.8 Flash (high, 23) and Claude Sonnet 5.5 (medium, 19). Its hallucination rate falls to 27% from Inkling Small’s 61%, while its attempt rate falls to 53% from 77% and accuracy rises to 34% from 32%.

Inkling 2 Small averages ~150K output tokens per Intelligence Index task, ~4x that of Inkling Small’s ~37K.

It also uses more output tokens than Gemini 3.8 Flash (high, ~99K) and Claude Sonnet 5.5 (medium, ~31K).

Inkling 2 Small costs $1.18 per Task in the Intelligence Index, up almost 6x from Inkling Small.

Token prices for Inkling 2 Small are $0.30/$1.00/$0.03 per 1M input/output/cache hit tokens.

Full results across the evaluations in the Artificial Analysis Intelligence Index v5.0.1