Compare / Qwen3 Next 80B A3B Thinking vs Llama 3.3 70B Instruct
Qwen3 Next 80B A3B Thinking 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.
Qwen3 Next 80B A3B Thinking or Llama 3.3 70B Instruct? The short answer
- Qwen3 Next 80B A3B Thinking is cheaper: $0.41 against $0.64 per million tokens, 36% less.
- Qwen3 Next 80B A3B Thinking scores higher on the Artificial Analysis Intelligence Index: 11.2 against 7.7.
- Qwen3 Next 80B A3B Thinking writes faster: 182 tokens a second against 80.
- On our business benchmarks, Qwen3 Next 80B A3B Thinking ranks higher for reasoning and knowledge and writing.
- On our business benchmarks, Llama 3.3 70B Instruct ranks higher for agents and tool use.
At a glance
| Qwen3 Next 80B A3B Thinking | Llama 3.3 70B Instruct | |
|---|---|---|
| Price per million tokens | $0.15 in · $1.20 out | $0.59 in · $0.79 out |
| Blended (3 in : 1 out) | $0.41 | $0.64 |
| Cheapest host, blended | $0.41 (Google) | $0.16 (DeepInfra, FP8) |
| Intelligence Index | 11.2 | 7.7 |
| Spring Prompt overall | – | – |
| Output speed, tokens a second | 182 | 80 |
| Context window | 262,144 tokens | 131,072 tokens |
| Weights | Open | Open |
| Developer based in | China | 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 | Qwen3 Next 80B A3B Thinking | Llama 3.3 70B Instruct |
|---|---|---|
| Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls | 137 of 161 | 127 of 161 |
| Reasoning and knowledgeHard questions across science, maths and general knowledge | 122 of 195 | 174 of 195 |
| WritingWhat people prefer in blind comparisons, and judged writing quality | 142 of 187 | 165 of 187 |
| Sticking to the factsSummarising without inventing things, and factual answers | 64 of 153 | not measured |
Every shared result
| Benchmark | Metric | Qwen3 Next 80B A3B Thinking | Llama 3.3 70B Instruct |
|---|---|---|---|
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,367 | 1,304 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,326 | 1,285 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,394 | 1,298 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,359 | 1,293 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,369 | 1,318 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,341 | 1,292 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 17.4 | 11.9 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 11.2 | 7.70 |
| Artificial Analysisreported | GPQA Diamond% of questions | 75.9% | 49.8% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 12.6% | 3.6% |
| Artificial Analysisreported | IFBench% of instructions | 60.7% | 47.1% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 63.7% | 15.7% |
| Artificial Analysisreported | Output speedtokens per second | 182 | 80.5 |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 6.7% | 4.9% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 9.8% | 3.0% |
| Artificial Analysisreported | Time to first answer tokenseconds | 12.1 | 0.64 |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 41.5% | 26.6% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 67.0% | 10.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.28 | 0.37 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 27.2 | 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 Qwen3 Next 80B A3B Thinking better than Llama 3.3 70B Instruct?
It depends on the task. Among models available today, on our benchmarks and licensed sources, Qwen3 Next 80B A3B Thinking ranks higher for reasoning and knowledge and writing; Llama 3.3 70B Instruct ranks higher for agents and tool use.
Which is cheaper, Qwen3 Next 80B A3B Thinking or Llama 3.3 70B Instruct?
Qwen3 Next 80B A3B Thinking costs $0.41 per million tokens (three input to one output), against $0.64 for Llama 3.3 70B Instruct.
Which is faster, Qwen3 Next 80B A3B Thinking or Llama 3.3 70B Instruct?
Qwen3 Next 80B A3B Thinking writes about 182 tokens a second on its usual API, against 80 for Llama 3.3 70B Instruct (Artificial Analysis).
Run Qwen3 Next 80B A3B Thinking 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.