Benchmark Almanac

BlueBench

GPT-4.1 64.6 leads since September 2025
Not in model cards

Agents solve enterprise tasks across knowledge, coding, security and business workflows. The benchmark aggregates capability scores from multiple public task suites.

IBM Research publishes an Apache 2.0 leaderboard space and describes BlueBench as an open benchmark developed with domain experts for enterprise use cases.

Made by
IBM Research
Released
September 2025
Kind
General assistants
Level
Expert
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Code → Text, Code, Actions
Interaction
Agentic
Environment
APIs and tools
Items from
Written by experts
Refresh
One fixed set
Run by
Anyone
Occupations
Computer and mathematical
Strongest model tested
o4-mini #95 on the capability index
Last model added
April 2025
Scores refreshed
November 21st, 2026
18
Entries on its own leaderboard, 13 models we track

GPT-4.1 leads with 64.6 among models released by April 2025 GPT-4.1 leads with 64.6 among models released by April 2025

Best score per entry from the benchmark's own leaderboard, read on November 21st, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

The newest model with a score here came out in April 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-4.1 64.6
  2. 2 GPT-4o Mar 2025 64.1
  3. 3 GPT-4.1 mini 63.0
  4. 4 meta-llama/llama-3-405b-instruct 61.6
  5. 5 mistralai/mistral-medium-2505 60.5
  6. 6 Llama 3.3 70B Instruct 59.0
  7. 7 o1 58.6
  8. 8 GPT-4.1 nano 57.0
  9. 9 Mistral Large 55.5
  10. 10 mistralai/mistral-small-3-1-24b-instruct-2503 54.1
  11. 11 o4-mini 54.0
  12. 12 o3-mini 53.9
  13. 13 Granite-3.0-8B-Instruct 49.7
  14. 14 Granite 3.3 8B Instruct 49.3
  15. 15 Pixtral 12B 45.4
  16. 16 meta-llama/llama-3-2-3b-instruct 44.3
  17. 17 Granite-3.0-2B-Instruct 41.0
  18. 18 meta-llama/llama-3-2-1b-instruct 31.8

Source: Official BlueBench leaderboard

#EntryScoreSettingDateSource
1 GPT-4.1 64.6 Apr 2025 Official leaderboard
2 GPT-4o Mar 2025 64.1 Mar 2025 Official leaderboard
3 GPT-4.1 mini 63.0 Apr 2025 Official leaderboard
4 meta-llama/llama-3-405b-instruct 61.6 Official leaderboard
5 mistralai/mistral-medium-2505 60.5 Official leaderboard
6 Llama 3.3 70B Instruct 59.0 Dec 2024 Official leaderboard
7 o1 58.6 Dec 2024 Official leaderboard
8 GPT-4.1 nano 57.0 Apr 2025 Official leaderboard
9 Mistral Large 55.5 Feb 2024 Official leaderboard
10 mistralai/mistral-small-3-1-24b-instruct-2503 54.1 Official leaderboard
11 o4-mini 54.0 Apr 2025 Official leaderboard
12 o3-mini 53.9 Jan 2025 Official leaderboard
13 Granite-3.0-8B-Instruct 49.7 Oct 2024 Official leaderboard
14 Granite 3.3 8B Instruct 49.3 Apr 2025 Official leaderboard
15 Pixtral 12B 45.4 Sep 2024 Official leaderboard
16 meta-llama/llama-3-2-3b-instruct 44.3 Official leaderboard
17 Granite-3.0-2B-Instruct 41.0 Oct 2024 Official leaderboard
18 meta-llama/llama-3-2-1b-instruct 31.8 Official leaderboard
#ModelScoreSettingDateSource
1 GPT-4.1 64.6 Apr 2025 Official leaderboard
2 GPT-4o Mar 2025 64.1 Mar 2025 Official leaderboard
3 GPT-4.1 mini 63.0 Apr 2025 Official leaderboard
4 Llama 3.3 70B Instruct 59.0 Dec 2024 Official leaderboard
5 o1 58.6 Dec 2024 Official leaderboard
6 GPT-4.1 nano 57.0 Apr 2025 Official leaderboard
7 Mistral Large 55.5 Feb 2024 Official leaderboard
8 o4-mini 54.0 Apr 2025 Official leaderboard
9 o3-mini 53.9 Jan 2025 Official leaderboard
10 Granite-3.0-8B-Instruct 49.7 Oct 2024 Official leaderboard
11 Granite 3.3 8B Instruct 49.3 Apr 2025 Official leaderboard
12 Pixtral 12B 45.4 Sep 2024 Official leaderboard
13 Granite-3.0-2B-Instruct 41.0 Oct 2024 Official leaderboard

Rankings by part

We read 13 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Bias, Chatbot Abilities and Entity Extraction. o1 leads 3 of them, and 5 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Bias

13 models we track have a score here.

  1. 1 o3-mini ↑8 98.0
  2. 2 GPT-4.1 ↓1 98.0
  3. 3 Mistral Large ↑4 97.0
  4. 4 Llama 3.3 70B Instruct 96.0
  5. 5 o1 94.9
  6. 6 GPT-4.1 mini ↓3 94.9
  7. 7 o4-mini ↑1 94.9
  8. 8 GPT-4o Mar 2025 ↓6 93.9
  9. 9 Granite-3.0-8B-Instruct ↑1 83.8
  10. 10 GPT-4.1 nano ↓4 81.8

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Own leaderboard Not tested since Sep 2025
GPT-4.1 64.6

Source: Official BlueBench leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard
40.0 45.0 50.0 55.0 60.0 65.0 2024 2025 40.0 45.0 50.0 55.0 60.0 65.0 2024 2025

Source: Official BlueBench leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Tool use

All 265 in Tool use
  • AutomationBench In use Agent completes end-to-end business workflows inside a simulated company exposing dozens of REST APIs. Graded on resulting state. Gemini 4 Neon 56.3%
  • MCP Atlas In use Multi-step workflows over real MCP servers. Scores tool-use competency on realistic tasks. Gemini 4 Argon 90.7%
  • BFCL V4 In use Models call functions and tools, including multi-turn agentic tasks with web search and memory. Checked against expected calls and outcomes. Qwen3.7-Max 75%
  • τ³-bench Banking In use Agents act as banking customer-service reps over many turns with a simulated user, following policy. Pass^1 on the final database state. Qwen 3.8 Max 55.1%
  • Toolathlon-Verified In use 108 long-horizon tasks across 32 apps through about 600 MCP tools. Execution-based checkers inspect final environment state. GLM-5.3-Flash 78.4%
  • DeepSearchQA In use Questions that need multi-step search and synthesis. Final answers graded against references. Kimi K3 95%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 19 months ago, in April 2025, for Granite 3.3 8B Instruct. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (18 entries).
  • Pass Models scored 13 models we track, plus 5 other entries on its leaderboard.
  • Warning Newest score The newest came 19 months ago, in April 2025, for Granite 3.3 8B Instruct.
  • Note Frontier models None of the top 10 has a score. The strongest tested is o4-mini (#95).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on huggingface.co: 13 rows for models we track and 5 other entries at the last read.
  • Pass Last read November 21st, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check