Compare / Llama 3.3 70B Instruct vs Mistral Large 3 2512
Llama 3.3 70B Instruct vs Mistral Large 3 2512
Which is cheaper, which is more capable, and which does better at each kind of work, from 20 results on 3 sources that measured both models.
Llama 3.3 70B Instruct or Mistral Large 3 2512? The short answer
- Llama 3.3 70B Instruct is cheaper: $0.64 against $0.75 per million tokens, 15% less.
- Mistral Large 3 2512 scores higher on the Artificial Analysis Intelligence Index: 9.3 against 7.7.
- On our business benchmarks, Llama 3.3 70B Instruct ranks higher for agents and tool use.
- On our business benchmarks, Mistral Large 3 2512 ranks higher for reasoning and knowledge and writing.
At a glance
| Llama 3.3 70B Instruct | Mistral Large 3 2512 | |
|---|---|---|
| Price per million tokens | $0.59 in · $0.79 out | $0.50 in · $1.50 out |
| Blended (3 in : 1 out) | $0.64 | $0.75 |
| Cheapest host, blended | $0.16 (DeepInfra, FP8) | $0.75 (Mistral) |
| Intelligence Index | 7.7 | 9.3 |
| Spring Prompt overall | – | – |
| Output speed, tokens a second | 80 | 80 |
| Context window | 131,072 tokens | 262,144 tokens |
| Weights | Open | Closed |
| Developer based in | the US | France |
Prices are the developer's list price where we have read it, otherwise the typical host on OpenRouter; see every model's price. Intelligence and speed from Artificial Analysis.
Which is better for what
Each model's rank among models available today, for each kind of work, from our own benchmarks and licensed sources. The better rank is in green.
| Use case | Llama 3.3 70B Instruct | Mistral Large 3 2512 |
|---|---|---|
| Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls | 127 of 161 | 131 of 161 |
| Reasoning and knowledgeHard questions across science, maths and general knowledge | 174 of 195 | 155 of 195 |
| WritingWhat people prefer in blind comparisons, and judged writing quality | 165 of 187 | 97 of 187 |
| Sticking to the factsSummarising without inventing things, and factual answers | not measured | 120 of 153 |
Every shared result
| Benchmark | Metric | Llama 3.3 70B Instruct | Mistral Large 3 2512 |
|---|---|---|---|
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,304 | 1,409 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,285 | 1,373 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,298 | 1,428 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,293 | 1,400 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,318 | 1,413 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,292 | 1,388 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 11.9 | 20.1 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 7.70 | 9.30 |
| Artificial Analysisreported | GPQA Diamond% of questions | 49.8% | 68.0% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 3.6% | 4.2% |
| Artificial Analysisreported | IFBench% of instructions | 47.1% | 36.2% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 15.7% | 36.0% |
| Artificial Analysisreported | Output speedtokens per second | 80.5 | 80.3 |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 4.9% | 12.0% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 3.0% | 15.9% |
| Artificial Analysisreported | Time to first answer tokenseconds | 0.64 | 0.69 |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 26.6% | 24.6% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 10.0% | 15.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.37 | 0.38 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 26.2 | 41.6 |
Bold green marks the better value on that metric. Where a source reports ranges that overlap, the difference may not be meaningful; see the benchmark page for ranges.
Quick answers
Is Llama 3.3 70B Instruct better than Mistral Large 3 2512?
It depends on the task. Among models available today, on our benchmarks and licensed sources, Llama 3.3 70B Instruct ranks higher for agents and tool use; Mistral Large 3 2512 ranks higher for reasoning and knowledge and writing.
Which is cheaper, Llama 3.3 70B Instruct or Mistral Large 3 2512?
Llama 3.3 70B Instruct costs $0.64 per million tokens (three input to one output), against $0.75 for Mistral Large 3 2512.
Which is faster, Llama 3.3 70B Instruct or Mistral Large 3 2512?
Llama 3.3 70B Instruct writes about 80 tokens a second on its usual API, against 80 for Mistral Large 3 2512 (Artificial Analysis).
Run Llama 3.3 70B Instruct and Mistral Large 3 2512 on your own prompt
Benchmarks aren't your data. Try both side by side in the playground, with the cost of every answer.