Inkling vs Inkling-Small: Specs & Benchmark Comparison

Inkling and Inkling-Small both come from Thinking Machines Lab, and both were released in July 2026. Inkling is the larger model at roughly 975 billion parameters, against 276 billion for Inkling-Small.

The two models share 13 published benchmarks. Inkling leads on 7 of them, Inkling-Small on 6. The widest gaps are on SimpleQA Verified, where Inkling scores 43.9% against 20.6%; Tau3 Banking, where Inkling scores 23.7% against 15.5%. Averaged across everything we track, Inkling sits at 61.8% and Inkling-Small at 60.3%.

Inkling-Small is the cheaper API at $0.30 per million input tokens and $1.20 per million output tokens, roughly 3 times cheaper than Inkling at $0.95 and $4.05. Inkling takes the larger context window at 524K tokens, compared with 256K for Inkling-Small. Both accept text, images, and audio as input.

CharacteristicInklingInkling-Small
CompanyThinking Machines LabThinking Machines Lab
Release DateJuly 21, 2026July 30, 2026
Parameters975B276B
MultimodalYesYes
Context (input)524K256K
Context (output)524K256K
Input Price / 1M$0.95$0.30
Output Price / 1M$4.05$1.20
Average Score61.8%60.3%
Benchmarks
SimpleQA Verified43.9%20.6%
Tau3 Banking23.7%15.5%
MCP Atlas76.0%79.6%
SWE-Bench Verified77.6%80.2%
IFBench79.8%82.2%
Global-MMLU-Lite88.7%86.7%
AIME 202697.1%95.5%
VoiceBench Avg91.4%90.1%
GDPval-AA41.3%42.3%
Terminal-Bench 2.163.8%64.7%
CharXiv-R78.1%77.4%
MMMU-Pro73.5%74.0%
MMAU77.2%77.0%

Visual Benchmark Comparison

Inkling
Inkling-Small
SimpleQA Verified0.4 vs 0.2
0.4
0.2
Tau3 Banking0.2 vs 0.2
0.2
0.2
MCP Atlas0.8 vs 0.8
0.8
0.8
SWE-Bench Verified0.8 vs 0.8
0.8
0.8
IFBench0.8 vs 0.8
0.8
0.8
Global-MMLU-Lite0.9 vs 0.9
0.9
0.9
AIME 20261.0 vs 1.0
1.0
1.0
VoiceBench Avg0.9 vs 0.9
0.9
0.9
GDPval-AA0.4 vs 0.4
0.4
0.4
Terminal-Bench 2.10.6 vs 0.6
0.6
0.6
CharXiv-R0.8 vs 0.8
0.8
0.8
MMMU-Pro0.7 vs 0.7
0.7
0.7
MMAU0.8 vs 0.8
0.8
0.8

Verdict

Both models show equal results — the choice depends on your specific use case.

Overall Performance

Both models show comparable average scores: Inkling — 0.6, Inkling-Small — 0.6.

Programming

On SWE-Bench, both models are nearly equal: Inkling — 0.8, Inkling-Small — 0.8.

API Cost

Inkling-Small is 3.3x cheaper: input $0.30/1M vs $0.95/1M tokens.

Context Window

Inkling supports a larger context: 524K vs 256K tokens.

Recency

Both models were released around the same time: 7/21/2026 and 7/30/2026.

More About These Models

Related Comparisons

Frequently Asked Questions

Which is better for coding — Inkling or Inkling-Small?
On the SWE-Bench benchmark, Inkling-Small shows a better result: 80.2% vs 77.6%.
Which model is cheaper — Inkling or Inkling-Small?
Inkling-Small is cheaper for input: $0.30 per 1M tokens vs $0.95.
Which has a larger context window — Inkling or Inkling-Small?
Inkling supports a larger context: 524,288 tokens vs 256,000.

The Inkling and Inkling-Small comparison is updated for 2026. Data includes benchmark results, API pricing, context window size and other specifications. For more detailed information, visit the Inkling or Inkling-Small page. See also the complete list of AI model comparisons.