Compare / Qwen3 8B vs Llama 3.3 70B Instruct
Qwen3 8B 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 8B or Llama 3.3 70B Instruct? The short answer
- Qwen3 8B is cheaper: $0.20 against $0.64 per million tokens, 3.2 times less.
- They score about the same on the Artificial Analysis Intelligence Index (7.3 and 7.7).
- On our business benchmarks, Qwen3 8B ranks higher for agents and tool use, reasoning and knowledge and writing.
At a glance
| Qwen3 8B | Llama 3.3 70B Instruct | |
|---|---|---|
| Price per million tokens | $0.12 in · $0.46 out | $0.59 in · $0.79 out |
| Blended (3 in : 1 out) | $0.20 | $0.64 |
| Cheapest host, blended | – | $0.16 (DeepInfra, FP8) |
| Intelligence Index | 7.3 | 7.7 |
| Spring Prompt overall | – | – |
| Output speed, tokens a second | – | 80 |
| Context window | 131,072 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 8B | Llama 3.3 70B Instruct |
|---|---|---|
| Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls | 126 of 161 | 127 of 161 |
| Reasoning and knowledgeHard questions across science, maths and general knowledge | 166 of 195 | 174 of 195 |
| WritingWhat people prefer in blind comparisons, and judged writing quality | 158 of 187 | 165 of 187 |
| Sticking to the factsSummarising without inventing things, and factual answers | 7 of 153 | not measured |
Every shared result
| Benchmark | Metric | Qwen3 8B | Llama 3.3 70B Instruct |
|---|---|---|---|
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 9.00 | 11.9 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 7.30 | 7.70 |
| Artificial Analysisreported | GPQA Diamond% of questions | 58.9% | 49.8% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 3.9% | 3.6% |
| Artificial Analysisreported | IFBench% of instructions | 33.5% | 47.1% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 0.0% | 15.7% |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 2.2% | 4.9% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 2.3% | 3.0% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 27.8% | 26.6% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Irrelevance detection% correct | 82.3% | 53.5% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Memory% correct | 14.6% | 8.2% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Multi-turn tasks% correct | 41.8% | 21.5% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Overall accuracy% correct | 42.6% | 31.9% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Relevance detection% correct | 93.8% | 100.0% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (curated)% correct | 88.6% | 88.0% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Single-turn calls (user-contributed)% correct | 80.5% | 76.6% |
| Berkeley Function Calling Leaderboard (BFCL) V4reported | Web search% correct | 13.5% | 10.0% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 17.0% | 10.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.36 | 0.37 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 28.0 | 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 8B better than Llama 3.3 70B Instruct?
It depends on the task. Among models available today, on our benchmarks and licensed sources, Qwen3 8B ranks higher for agents and tool use, reasoning and knowledge and writing.
Which is cheaper, Qwen3 8B or Llama 3.3 70B Instruct?
Qwen3 8B costs $0.20 per million tokens (three input to one output), against $0.64 for Llama 3.3 70B Instruct.
Run Qwen3 8B 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.