Compare / Gemini 3.1 Pro Preview vs Llama 3.3 70B Instruct
Gemini 3.1 Pro Preview vs Llama 3.3 70B Instruct
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.
Gemini 3.1 Pro Preview or Llama 3.3 70B Instruct? The short answer
- Llama 3.3 70B Instruct is cheaper: $0.64 against $4.50 per million tokens, 7.0 times less.
- Gemini 3.1 Pro Preview scores higher on the Artificial Analysis Intelligence Index: 29.7 against 7.7.
- Gemini 3.1 Pro Preview writes faster: 121 tokens a second against 80.
- On our business benchmarks, Gemini 3.1 Pro Preview ranks higher for agents and tool use, reasoning and knowledge and writing.
At a glance
| Gemini 3.1 Pro Preview | Llama 3.3 70B Instruct | |
|---|---|---|
| Price per million tokens | $2.00 in · $12.00 out | $0.59 in · $0.79 out |
| Blended (3 in : 1 out) | $4.50 | $0.64 |
| Cheapest host, blended | $2.25 (Google) | $0.16 (DeepInfra, FP8) |
| Intelligence Index | 29.7 | 7.7 |
| Spring Prompt overall | 47 | – |
| Output speed, tokens a second | 121 | 80 |
| Context window | 1,048,576 tokens | 131,072 tokens |
| Weights | Closed | Open |
| Developer based in | the US | the US |
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 | Gemini 3.1 Pro Preview | Llama 3.3 70B Instruct |
|---|---|---|
| Product listingsTurning a sparse product feed and photos into listings that can go live | 8 of 19 | not measured |
| Decks from an analysisTurning a finished analysis into a deck you could present as it is | 12 of 18 | not measured |
| Marketing planningPlanning a year of ad spend without overspending | 9 of 18 | not measured |
| Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls | 60 of 161 | 127 of 161 |
| Professional workReal tasks from banking, consulting and law, business documents and freelance projects | 36 of 51 | not measured |
| Reasoning and knowledgeHard questions across science, maths and general knowledge | 23 of 195 | 174 of 195 |
| WritingWhat people prefer in blind comparisons, and judged writing quality | 9 of 187 | 165 of 187 |
| Sticking to the factsSummarising without inventing things, and factual answers | 35 of 153 | not measured |
| Speed under pressureGood decisions against a real clock (fast chess) | 5 of 21 | not measured |
Every shared result
| Benchmark | Metric | Gemini 3.1 Pro Preview | Llama 3.3 70B Instruct |
|---|---|---|---|
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,477 | 1,304 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,480 | 1,285 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,509 | 1,298 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,479 | 1,293 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,487 | 1,318 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,480 | 1,292 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 68.8 | 11.9 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 29.7 | 7.70 |
| Artificial Analysisreported | GPQA Diamond% of questions | 94.1% | 49.8% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 47.0% | 3.6% |
| Artificial Analysisreported | IFBench% of instructions | 77.1% | 47.1% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 82.0% | 15.7% |
| Artificial Analysisreported | Output speedtokens per second | 121 | 80.5 |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 73.8% | 4.9% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 53.8% | 3.0% |
| Artificial Analysisreported | Time to first answer tokenseconds | 15.2 | 0.64 |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 95.6% | 26.6% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 19.0% | 10.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.34 | 0.37 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 72.2 | 26.2 |
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 Gemini 3.1 Pro Preview better than Llama 3.3 70B Instruct?
It depends on the task. Among models available today, on our benchmarks and licensed sources, Gemini 3.1 Pro Preview ranks higher for agents and tool use, reasoning and knowledge and writing.
Which is cheaper, Gemini 3.1 Pro Preview or Llama 3.3 70B Instruct?
Llama 3.3 70B Instruct costs $0.64 per million tokens (three input to one output), against $4.50 for Gemini 3.1 Pro Preview.
Which is faster, Gemini 3.1 Pro Preview or Llama 3.3 70B Instruct?
Gemini 3.1 Pro Preview writes about 121 tokens a second on its usual API, against 80 for Llama 3.3 70B Instruct (Artificial Analysis).
Run Gemini 3.1 Pro Preview and Llama 3.3 70B Instruct on your own prompt
Benchmarks aren't your data. Try both side by side in the playground, with the cost of every answer.