Compare / DeepSeek V3.2 vs Llama 3.1 8B Instruct
DeepSeek V3.2 vs Llama 3.1 8B Instruct
Which is cheaper, which is more capable, and which does better at each kind of work, from 18 results on 3 sources that measured both models.
DeepSeek V3.2 or Llama 3.1 8B Instruct? The short answer
- Llama 3.1 8B Instruct is cheaper: $0.058 against $0.46 per million tokens, 8.1 times less.
- DeepSeek V3.2 scores higher on the Artificial Analysis Intelligence Index: 21.5 against 6.9.
- On our business benchmarks, DeepSeek V3.2 ranks higher for agents and tool use, reasoning and knowledge and writing.
At a glance
| DeepSeek V3.2 | Llama 3.1 8B Instruct | |
|---|---|---|
| Price per million tokens | $0.30 in · $0.96 out | $0.050 in · $0.080 out |
| Blended (3 in : 1 out) | $0.46 | $0.058 |
| Cheapest host, blended | $0.23 (GMICloud, FP8) | $0.025 (DeepInfra, FP8) |
| Intelligence Index | 21.5 | 6.9 |
| Spring Prompt overall | – | – |
| Output speed, tokens a second | – | – |
| Context window | 163,840 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 | DeepSeek V3.2 | Llama 3.1 8B Instruct |
|---|---|---|
| Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls | 83 of 161 | 154 of 161 |
| Reasoning and knowledgeHard questions across science, maths and general knowledge | 81 of 195 | 173 of 195 |
| WritingWhat people prefer in blind comparisons, and judged writing quality | 64 of 187 | 178 of 187 |
| Sticking to the factsSummarising without inventing things, and factual answers | 56 of 153 | not measured |
Every shared result
| Benchmark | Metric | DeepSeek V3.2 | Llama 3.1 8B Instruct |
|---|---|---|---|
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,423 | 1,210 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,400 | 1,177 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,453 | 1,198 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,420 | 1,191 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,425 | 1,211 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,410 | 1,182 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 44.2 | 5.40 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 21.5 | 6.90 |
| Artificial Analysisreported | GPQA Diamond% of questions | 84.0% | 25.9% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 24.6% | 5.3% |
| Artificial Analysisreported | IFBench% of instructions | 60.7% | 28.6% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 73.3% | 18.0% |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 46.8% | 1.5% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 35.6% | 0.8% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 90.6% | 16.4% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 26.0% | 20.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.36 | 0.34 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 57.3 | 24.8 |
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 DeepSeek V3.2 better than Llama 3.1 8B Instruct?
It depends on the task. Among models available today, on our benchmarks and licensed sources, DeepSeek V3.2 ranks higher for agents and tool use, reasoning and knowledge and writing.
Which is cheaper, DeepSeek V3.2 or Llama 3.1 8B Instruct?
Llama 3.1 8B Instruct costs $0.058 per million tokens (three input to one output), against $0.46 for DeepSeek V3.2.
Run DeepSeek V3.2 and Llama 3.1 8B Instruct on your own prompt
Benchmarks aren't your data. Try both side by side in the playground, with the cost of every answer.