Scale Labs FORTRESS (FORTRESS)

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 FORTRESS — November 12, 2026 snapshot

We mirror the published scale score view for FORTRESS. DeepSeek-R1 leads the public snapshot at 74.4%, followed by glm-4p5-air (63.2%) and Gemini 2.5 Pro (06-05) (61.7%). We do not use these results to rank models overall.

1Open weights

DeepSeek-R1

DeepSeek

74.4%

Overall 40.66Context 128K

glm-4p5-air

Z.AI

63.2%

Gemini 2.5 Pro (06-05)

Google

61.7%

82 modelsKnowledgeCurrentDisplay onlyUpdated November 12, 2026 snapshot

Scale score table (82 models)

Scale score results for FORTRESS
Rank	Model / configuration	Score	Parameters (B)	Open / closed
1	DeepSeek-R1DeepSeek	74.4%	Not reported	Open
2	glm-4p5-airZ.AI	63.2%	Not reported	Unknown
3	Gemini 2.5 Pro (06-05)Google	61.7%	Not reported	Unknown
4	Qwen3 235B-A22BAlibaba	61.4%	Not reported	Open
5	DeepSeek V3.1DeepSeek	60.5%	Not reported	Open
6	glm-4p5Z.AI	59.6%	Not reported	Unknown
7	GPT-4.1 miniOpenAI	59.2%	Not reported	Closed
8	Qwen 2.5 72BAlibaba	56.4%	Not reported	Unknown
9	Mixtral 8x22BMistral	56.1%	Not reported	Unknown
10	kimi-k2-instructMoonshot AI	55.5%	Not reported	Unknown
11	Gemini 2.5 Pro (03-25)Google	54.9%	Not reported	Unknown
12	Gemini 1.5 ProGoogle	53.9%	Not reported	Closed
13	GPT-4.1OpenAI	53.0%	Not reported	Closed
14	minimax-m3p1-flashMiniMax	52.1%	Not reported	Unknown
15	gemini-3.1-flash-lite-previewGoogle	51.1%	Not reported	Unknown
16	Gemini 1.5 FlashGoogle	50.6%	Not reported	Closed
17	GLM-5.2Z.AI	48.7%	Not reported	Open
18	GPT-4o miniOpenAI	48.1%	Not reported	Closed
19	GPT-4oOpenAI	47.2%	Not reported	Closed
20	Llama 3.3 70BMeta	44.8%	Not reported	Open
21	Llama 3.1 70BMeta	44.2%	Not reported	Open
22	gemini-3-pro-previewGoogle	41.7%	Not reported	Unknown
23	Kimi K2.5Moonshot AI	41.1%	Not reported	Open
24	Llama 4 MaverickMeta	40.1%	Not reported	Open
25	Claude 3.7 SonnetAnthropic	38.0%	Not reported	Closed
26	Claude 3.5 HaikuAnthropic	30.4%	Not reported	Closed
27	gpt-5.1-instantOpenAI	30.4%	Not reported	Unknown
28	o3-miniOpenAI	30.1%	Not reported	Closed
29	gemini-3.1-pro-previewGoogle	29.8%	Not reported	Unknown
30	glm-5p3 maxZ.AI	28.2%	Not reported	Unknown
31	Claude Opus 4Anthropic	27.6%	Not reported	Unknown
32	Kimi K3Moonshot AI	27.3%	Not reported	Open
33	deepseek-v4p1-flash-1017Deepseek	27.2%	Not reported	Unknown
34	glm-5p3-flashZ.AI	26.7%	Not reported	Unknown
35	kimi-k3 maxMoonshot AI	26.6%	Not reported	Unknown
36	nemotron-3-ultra-nvfp4 highNvidia	26.4%	Not reported	Unknown
37	gpt-5.1-thinkingOpenAI	25.7%	Not reported	Unknown
38	inkling xhighThinking Machines	25.0%	Not reported	Unknown
39	Claude 4 Opus (thinking)Anthropic	24.8%	Not reported	Unknown
40	Claude Sonnet 4Anthropic	24.4%	Not reported	Unknown
41	Mistral Large 4Mistral	23.1%	Not reported	Open
42	qwen3p8-2p4t-a95b xhighAlibaba	21.6%	Not reported	Unknown
43	o4-miniOpenAI	21.5%	Not reported	Closed
44	Llama 3.1 405BMeta	20.6%	Not reported	Open
45	claude-opus-4-6 (Non-Thinking)Anthropic	20.5%	Not reported	Unknown
46	Muse SparkMeta	20.2%	Not reported	Closed
47	GPT-5.6 SolOpenAI	20.1%	Not reported	Closed
48	o1OpenAI	19.4%	Not reported	Closed
49	kimi-k3p1-proMoonshot AI	18.6%	Not reported	Unknown
50	Claude Opus 4.8 maxAnthropic	18.2%	Not reported	Unknown
51	GPT-6.1 TerraOpenAI	18.2%	Not reported	Closed
52	Claude 4 Sonnet (thinking)Anthropic	18.1%	Not reported	Unknown
53	GPT-OSS 20BOpenAI	17.6%	Not reported	Open
54	GPT-5.2OpenAI	17.5%	Not reported	Closed
55	gpt-5-2025-08-07OpenAI	17.0%	Not reported	Unknown
56	GPT-5 miniOpenAI	17%	Not reported	Closed
57	Claude Sonnet 4.5Anthropic	17.0%	Not reported	Closed
58	Claude Opus 5Anthropic	16.8%	Not reported	Closed
59	Gemini 4 CarbonGoogle	16.4%	Not reported	Closed
60	gpt-5.5-xhighOpenAI	16.3%	Not reported	Unknown
61	claude-opus-4-1-20250805Anthropic	16.1%	Not reported	Unknown
62	o3OpenAI	16.0%	Not reported	Closed
63	Claude Fable 5.2Anthropic	15.6%	Not reported	Closed
64	gpt-5-pro-2025-10-06OpenAI	15.2%	Not reported	Unknown
65	GPT-5.4 ProOpenAI	14.8%	Not reported	Closed
66	claude-opus-4-1-20250805-thinkingAnthropic	14.8%	Not reported	Unknown
67	Fable-5.1Anthropic	13.7%	Not reported	Unknown
68	Claude Opus 4.5Anthropic	13.6%	Not reported	Closed
69	GPT-6 SolOpenAI	13.5%	Not reported	Closed
70	Claude Opus 5.5Anthropic	13.2%	Not reported	Closed
71	claude-opus-4-6-thinking-maxAnthropic	13%	Not reported	Unknown
72	Claude 3.5 SonnetAnthropic	13.0%	Not reported	Closed
73	Claude Sonnet 4.5 ThinkingAnthropic	12.8%	Not reported	Closed
74	Muse Spark 1.1Meta	12.4%	Not reported	Closed
75	GPT-6 AstraOpenAI	12.1%	Not reported	Closed
76	beam-1p0-501b-a23bReflection	11.9%	Not reported	Unknown
77	GPT-6 LunaOpenAI	10.8%	Not reported	Closed
78	Claude Opus 4.5 ThinkingAnthropic	9.6%	Not reported	Closed
79	Claude Haiku 5.5Anthropic	9.5%	Not reported	Closed
80	GPT-OSS 120BOpenAI	8.2%	Not reported	Open
81	Muse Spark OpenMeta	7.6%	Not reported	Open
82	claude-opus-5-5-maxAnthropic	5.8%	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: 82 sourced rows are currently displayable on this page; the leading published row is DeepSeek-R1 at 74.4%.

Honest limit: This Scale table is display-only context, not benchmark provenance or a weighted model-only comparison.

FORTRESS on Scale Labs

How BenchLM shows FORTRESS

BenchLM mirrors 82 published rows from Scale Labs’ public FORTRESS 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.

Snapshot

82 published rowsScale Labs sourceDisplay only

FORTRESS on Scale Labs

The published FORTRESS snapshot places DeepSeek-R1 first at 74.4%. The third row is 12.7 points behind. The broader top-10 range is 18.9 points, so the table still separates the published systems.

82 models have been evaluated on FORTRESS. 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. FORTRESS is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About FORTRESS

Year

2026

Tasks

82 published rows

Format

Published Scale leaderboard score

Difficulty

External agent and model evaluation

BenchLM mirrors 82 published rows from the FORTRESS public table captured on November 12, 2026 snapshot.

Scale Labs leaderboard Public benchmark source

Freshness and provenance

Version

FORTRESS 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

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 FORTRESS measure?

A Scale Labs public leaderboard mirrored as display-only reference data. It does not affect BenchLM rankings.

Which model leads the published FORTRESS snapshot?

DeepSeek-R1 currently leads the published FORTRESS snapshot with 74.4% scale score. BenchLM shows this benchmark for display only and does not use it in overall rankings.

How many models are evaluated on FORTRESS?

The November 12, 2026 snapshot snapshot contains 82 AI models.

Last updated: November 12, 2026 snapshot · mirrored from the public benchmark leaderboard

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 7,000+ subscribers.

One email each week. Unsubscribe anytime.