GPT-5.6 Sol vs Mistral Medium 3.5 128B
Wins 6 of 7 areas
Coding · Agents · Reasoning · Images and charts · Long documents · Following instructions
Wins 0 of 7 areas
—
GPT-5.6 Sol is the stronger all-rounder.Mistral Medium 3.5 128B is cheaper.
Scores updated · 23 tests both models report · How we compare
Where each one wins
Tests won in each of the seven areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- CodingWriting and fixing software40GPT-5.6 Sol4 of 4 tests
- AgentsCarrying out multi-step tasks on its own40GPT-5.6 Sol4 of 4 tests
- ReasoningHard problems that need careful thinking30GPT-5.6 Sol3 of 3 tests
- Images and chartsUnderstanding pictures, charts and video20GPT-5.6 Sol2 of 2 tests
- Long documentsFinding answers in very long texts10GPT-5.6 Sol1 of 1 test
- Following instructionsDoing exactly what it is asked10GPT-5.6 Sol1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
Long documents and following instructions rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
Mistral Medium 3.5 128B costs 63% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where GPT-5.6 Sol pulls ahead
- Real work tasks from 44 professionsGDPVal+42.4points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+41.8points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+37.4points ahead
Where Mistral Medium 3.5 128B pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+10.6points ahead
Every test, side by side
All 23 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5.6 Sol
- Terminal-Bench 2.1GPT-5.6 Sol by 37.48850.6+37.4
- LMArena · WebDevGPT-5.6 Sol by 353 rating points16181265+353 rating
- Terminal-Bench HardGPT-5.6 Sol by 32.665.933.3+32.6
- SciCodeGPT-5.6 Sol by 16.957.140.2+16.9
AgentsGPT-5.6 Sol
- GDPValGPT-5.6 Sol by 42.455.613.2+42.4
- BrowseCompGPT-5.6 Sol by 41.890.448.6+41.8
- Terminal-Bench 4.0GPT-5.6 Sol by 39.939.90+39.9
- τ-Bench V3 · BankingGPT-5.6 Sol by 29.244.315.1+29.2
ReasoningGPT-5.6 Sol
- Humanity's Last ExamGPT-5.6 Sol by 35.749.513.8+35.7
- CritPtGPT-5.6 Sol by 32.332.30+32.3
- GPQA DiamondGPT-5.6 Sol by 19.394.174.8+19.3
FactsEven
- AA-Omniscience · AccuracyGPT-5.6 Sol by 34.759.424.7+34.7
- AA-Omniscience · Non-hallucinationMistral Medium 3.5 128B by 10.67.818.4+10.6
Images and chartsGPT-5.6 Sol
- MMMU-ProGPT-5.6 Sol by 18.583.464.9+18.5
- LMArena · VisionGPT-5.6 Sol by 57 rating points12791222+57 rating
Following instructionsGPT-5.6 Sol
- IFBenchGPT-5.6 Sol by 3.972.768.8+3.9
Other results6 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- AA-OmniscienceGPT-5.6 Sol by 58.822-36.8+58.8
- AA Agentic IndexGPT-5.6 Sol by 41.150.59.4+41.1
- AA IntelligenceGPT-5.6 Sol by 32.84714.2+32.8
- Artificial Analysis Coding IndexGPT-5.6 Sol by 30.577.446.9+30.5
- τ³-Bench BankingGPT-5.6 Sol by 19.63313.4+19.6
- τ²-Bench Telecom (AA run)Mistral Medium 3.5 128B by 9.185.194.2+9.1
Questions people ask
Which is better, GPT-5.6 Sol or Mistral Medium 3.5 128B?
GPT-5.6 Sol wins six of the seven areas where both have results: coding, agents, reasoning, images and charts, long documents and following instructions. Mistral Medium 3.5 128B wins none, but costs 63% less. They are level on facts.
Which is better for coding?
GPT-5.6 Sol. It wins 4 of the 4 coding tests both models report; Mistral Medium 3.5 128B wins none.
Which is cheaper?
GPT-5.6 Sol costs $4.00 per million input tokens and $20.00 per million output tokens; Mistral Medium 3.5 128B costs $1.50 and $7.50. That makes Mistral Medium 3.5 128B about 63% cheaper for the same work.
How do you compare the two?
We use the 23 benchmark tests both models have published scores on. The verdict counts the 17 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 6 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.