DeepSeek V3 deepseek-v3
deepseek-ai · deepseek flagship
#22 overall #22 coding #12 agentic
- params
- 671B (A37B)
- arch
- moe
- context
- 128k
- license
- mit open weights
- released
- Dec 2024
- train compute
- 3.3e+24 FLOP
- downloads/30d
- 1.0M
# why ranked
Overall: #22 — score 14.2 ● low — a single signal; treat with caution
| signal | weight | input (0–100) |
|---|---|---|
| aa_intelligence | 0.50 | 14.2 |
| arena_elo | 0.30 | — |
| bench_composite | 0.20 | — |
missing signals are dropped and the remaining weights renormalized — never imputed.
Coding: #22 — score 37.1 ● high — all signals present
| signal | weight | input (0–100) |
|---|---|---|
| aa_coding | 0.25 | 50.0 |
| aider_polyglot | 0.25 | 54.5 |
| arena_elo_code | 0.15 | — |
| swe_bench_verified | 0.35 | 15.4 |
missing signals are dropped and the remaining weights renormalized — never imputed.
Agentic: #12 — score 15.4 ● low — a single signal; treat with caution
| signal | weight | input (0–100) |
|---|---|---|
| swe_bench_verified | 1.00 | 15.4 |
# trend
# benchmarks
| benchmark | score | source | date |
|---|---|---|---|
| AA Coding Index | 23.0 | Artificial Analysis | — |
| AA Intelligence Index | 14.2 | Artificial Analysis | — |
| AA Math Index | 26.0 | Artificial Analysis | — |
| Aider Polyglot | 55.1 | Aider | Mar 2025 |
| SWE-bench Verified | 42.0 | SWE-bench | Aug 2025 |
# where to run
| provider | quant | ctx | $/M in | $/M out | $/M blend | tps | uptime |
|---|---|---|---|---|---|---|---|
| StreamLake | unknown | 125k | $0.20 | $0.80 | $0.35 | — | 99.9% |
| DeepInfra | fp4 | 160k | $0.32 | $0.89 | $0.46 | — | 99.7% |
| Novita | fp8 | 64k | $0.40 | $1.30 | $0.62 | — | 99.6% |
blended = (3·input + output) / 4 per 1M tokens · cheapest first
# variants
source aliases
- aa
deepseek-v3- aider
DeepSeek Chat V3 (prev),DeepSeek V3 (0324)- arena
DeepSeek-V3- epoch
DeepSeek-V3 (Mar 2025)- litellm
deepseek/deepseek-chat- openrouter
deepseek/deepseek-chat- swebench
DeepSeek-V3-0324