Scale Labs DrugDiscoveryBench (DrugDiscoveryBench)
We show this table for reference; we do not rank on it.
A Scale Labs public leaderboard mirrored as display-only reference data. It does not affect BenchLM rankings.
Scale score on DrugDiscoveryBench — November 12, 2026 snapshot
We mirror the published scale score view for DrugDiscoveryBench. GPT 6 Astra (mini-SWE-agent) max leads the public snapshot at 68.7%, followed by GPT 6 Astra (Codex) max (65.4%) and Gemini 4 Argon (mini-swe-agent) high (65.4%). We do not use these results to rank models overall.
GPT 6 Astra (mini-SWE-agent) max
OpenAI
GPT 6 Astra (Codex) max
OpenAI
Gemini 4 Argon (mini-swe-agent) high
26 modelsKnowledgeCurrentDisplay onlyUpdated November 12, 2026 snapshot
Scale score table (26 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | GPT 6 Astra (mini-SWE-agent) maxOpenAI | 68.7% | Not reported | Unknown |
| 2 | GPT 6 Astra (Codex) maxOpenAI | 65.4% | Not reported | Unknown |
| 3 | Gemini 4 Argon (mini-swe-agent) highGoogle | 65.4% | Not reported | Unknown |
| 4 | Kimi K3p1 Pro (mini-swe-agent) xhighMoonshot AI | 63.4% | Not reported | Unknown |
| 5 | Muse-Spark 1.3 (mini-SWE-agent) highMeta | 62.2% | Not reported | Unknown |
| 6 | Deepseek V4p1 Flash 1017 (mini-swe-agent) xhighDeepseek | 60.6% | Not reported | Unknown |
| 7 | GPT 5.6 Sol (Codex) maxOpenAI | 56.5% | Not reported | Unknown |
| 8 | Claude Opus 5 (Claude Code) maxAnthropic | 53.7% | Not reported | Unknown |
| 9 | GPT-5.5 (mini-SWE-agent) xhighOpenAI | 51.6% | Not reported | Unknown |
| 10 | Claude Sonnet 5 (mini-swe-agent) maxAnthropic | 50.0% | Not reported | Unknown |
| 11 | Gemini 3.5 Flash (Gemini CLI) highGoogle | 50% | Not reported | Unknown |
| 12 | Gemini 3.5 Flash (mini-SWE-agent) highGoogle | 48.8% | Not reported | Unknown |
| 13 | Claude Opus 4.8 (mini-SWE-agent) maxAnthropic | 46.8% | Not reported | Unknown |
| 14 | Claude Opus 4.8 (Claude Code) maxAnthropic | 45.1% | Not reported | Unknown |
| 15 | GPT-5.5 (Codex) xhighOpenAI | 45.1% | Not reported | Unknown |
| 16 | Claude Sonnet 5.0 (Claude Code) maxAnthropic | 44.7% | Not reported | Unknown |
| 17 | Gemini 3.1 Pro (Gemini CLI) highOpenAI | 41.9% | Not reported | Unknown |
| 18 | GLM 5.2 (mini-SWE-agent) xhighZ.AI | 36.2% | Not reported | Unknown |
| 19 | Kimi K2.7 Code (mini-SWE-agent) xhighMoonshot AI | 35.3% | Not reported | Unknown |
| 20 | DeepSeek V4 Pro (mini-SWE-agent) xhighDeepseek | 31.7% | Not reported | Unknown |
| 21 | Claude Sonnet 4.6 (mini-SWE-agent) maxAnthropic | 31.3% | Not reported | Unknown |
| 22 | Qwen 3.7 Max (mini-SWE-agent) xhighQwen | 29.3% | Not reported | Unknown |
| 23 | GPT-5.2 (Codex) xhighOpenAI | 29.3% | Not reported | Unknown |
| 24 | Claude Opus 4.6 (Claude Code) maxAnthropic | 27.7% | Not reported | Unknown |
| 25 | Claude Sonnet 4.6 (Claude Code) maxAnthropic | 24% | Not reported | Unknown |
| 26 | MiniMax M3 (mini-SWE-agent) xhighMiniMax | 22.8% | Not reported | Unknown |
How to read this leaderboard
Compare the published configurations as complete evaluation systems. The source can combine a base model, agent scaffold, tools, budget, and inference setting in each result.
Operator receipt: 26 sourced rows are currently displayable on this page; the leading published row is GPT 6 Astra (mini-SWE-agent) max at 68.7%.
Honest limit: This Scale table is display-only context, not benchmark provenance or a weighted model-only comparison.
How BenchLM shows DrugDiscoveryBench
BenchLM mirrors 26 published rows from Scale Labs’ public DrugDiscoveryBench leaderboard, captured on November 12, 2026 snapshot.
The table is display only. It is useful context for a published agent or model configuration, but it does not enter BenchLM’s overall or category rankings.
The published DrugDiscoveryBench snapshot places GPT 6 Astra (mini-SWE-agent) max first at 68.7%. The third row is 3.3 points behind. The broader top-10 range is 18.7 points, so the table still separates the published systems.
26 models have been evaluated on DrugDiscoveryBench. The benchmark falls in the Knowledge category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. DrugDiscoveryBench is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About DrugDiscoveryBench
Year
2026
Tasks
26 published rows
Format
Published Scale leaderboard score
Difficulty
External agent and model evaluation
BenchLM mirrors 26 published rows from the DrugDiscoveryBench public table captured on November 12, 2026 snapshot.
Freshness and provenance
Version
DrugDiscoveryBench 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does DrugDiscoveryBench measure?
A Scale Labs public leaderboard mirrored as display-only reference data. It does not affect BenchLM rankings.
Which model leads the published DrugDiscoveryBench snapshot?
GPT 6 Astra (mini-SWE-agent) max currently leads the published DrugDiscoveryBench snapshot with 68.7% scale score. BenchLM shows this benchmark for display only and does not use it in overall rankings.
How many models are evaluated on DrugDiscoveryBench?
The November 12, 2026 snapshot snapshot contains 26 AI models.
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.