DeepSeek-V3.2-Exp · Aider coding benchmarks · Polyglot · percent correct ↗
1 more reported result
70.2% · DeepSeek-V3.2-Exp · Aider coding benchmarks · Polyglot · percent correct · Epoch AI · Aider polyglot ↗
DeepSeek / MODEL EVIDENCE
DeepSeek V3.2 Exp
12 benchmarks · 15 observations · 13 sources · released 2025-09-29
Scores at a glance
12 benchmarks · large dot: DeepSeek V3.2 Exp; small dots: closest models where reported
Fiction.liveBench ↗83.3%
DeepSeek-V3.2-Exp · Fiction.LiveBench · 16k token score · % ↗
1 more reported result
52.8% · DeepSeek-V3.2-Exp · Fiction.LiveBench · 16k token score · % · Epoch AI · Fiction.LiveBench ↗
TheAgentCompany ↗42.9%
DeepSeek-V3.2-Exp · The Agent Company · % Resolved · % ↗
WeirdML (v2) ↗39.48%
DeepSeek-V3.2-Exp · WeirdML · Accuracy · % ↗
CritPt ↗2.86%
DeepSeek-V3.2-Exp · CritPt · Accuracy · % ↗
CL-bench ↗13.2%
DeepSeek-V3.2-Exp · CL-bench · Overall · % ↗
CL-bench Life ↗9.5%
DeepSeek-V3.2-Exp · CL-bench Life · Overall · % ↗
LMCA ↗29.11%
DeepSeek-V3.2-Exp · LMCA · Score · % ↗
DTBench ↗87.73%
DeepSeek-V3.2-Exp · DTBench · Accuracy · % ↗
SciCode ↗38.89%
DeepSeek-V3.2-Exp · SciCode · Score · % ↗
Text Arena (Coding) ↗1271.75 Elo
DeepSeek-V3.2-Exp · WebDev Arena · Elo ↗
1 more reported result
1367 Elo · DeepSeek-V3.2-Exp · WebDev Arena · Elo · Epoch AI · WebDev Arena ↗
Epoch Capabilities Index (ECI) ↗144.91 ECI
DeepSeek-V3.2-Exp · General ECI · 26 October 2026 fit ↗
Closest models by score
Similar positions on shared benchmark leaderboards. Smaller gaps are closer.
Claude Sonnet 4 →Anthropic
6 of 9 shared benchmark comparisons
Mistral Medium 3.5 →Mistral
6 of 9 shared benchmark comparisons
6 of 9 shared benchmark comparisons
How similarity is measured
Mean absolute gap in percentile positions on shared benchmark settings; lower is closer. At least 60% overlap is required, with a minimum of three when available. Independent evaluations are used when there are at least three; otherwise published report comparisons are used. Repeated model observations within one setting are averaged before ranking. Writing-style traits and composite indices are excluded. This describes similarity, not an overall quality ranking.
Sources
Every plotted observation links to its evaluation. Reported configurations, benchmark versions and task subsets can differ; similarity describes shared scores, not training lift.
- Aider LLM Leaderboards — Polyglot coding benchmark ↗
- Arena leaderboard — WebDev evaluation ↗
- SciCode Benchmark | Artificial Analysis ↗
- DTBench — Conceptual Reasoning Index ↗
- LMCA — Conceptual Reasoning Index ↗
- CL-bench ↗
- CL-bench Life ↗
- CritPt v1 ↗
- The Agent Company ↗
- Text Arena (Coding) ↗
- AI benchmarks & capabilities ↗
- Fiction.liveBench — March 14, 2025 benchmark release ↗
- WeirdML ↗
Names used in source tables
DeepSeek-V3.2-Exp