GPT-5.6 Sol vs MiMo V2.5
29 SHARED BENCHMARKSAcross 29 shared benchmarks, GPT-5.6 Sol scores higher on 26 and MiMo V2.5 on 3. The widest gap is AA-Omniscience · Non-hallucination, where MiMo V2.5 scores 68.1 against 7.8. MiMo V2.5 is the cheaper of the two on tracked API pricing ($0.14 against $4.00 per million input tokens).
At a glance
Release dates per Artificial Analysis. Prices: OpenAI's own price page for GPT-5.6 Sol; Artificial Analysis for MiMo V2.5. ◆ = independently verified score.
Where each leadsby capability area · benchmark wins
Biggest gaps
GPT-5.6 Sol pulls furthest ahead on
- CritPt32.3 vs 3.7
- τ-Bench V3 · Banking44.3 vs 8.7
- SimpleQA Verified69.7 vs 16.1
MiMo V2.5 pulls furthest ahead on
- AA-Omniscience · Non-hallucination68.1 vs 7.8
Every shared benchmark29 · grouped by area
Reasoning 3
Coding 5
Agentic 3
Factuality 3
Instruction Following 1
Long Context 1
Math 1
Multimodal 3
Other shared benchmarks 9
Each score is the one the model's own page shows — the most authoritative tracked result for that benchmark, on the benchmark's standard methodology (source-attributed). ↓ marks lower-is-better metrics; ◆ an independently verified score. Open either model for its full surface, provenance and pricing. Quoted rates are the price-setter row we currently track for each model — its direct or vendor-official listing where one exists (openai-official, direct), otherwise the lowest tracked offer.