Benchmark Almanac

ToolRet

bge-reranker-v2-gemma 54.8 tops its leaderboard
Not in model cards

Models retrieve relevant tools for 7,600 requests from a corpus of about 43,000 tools. Evaluation measures ranking quality, precision, recall, and complete tool-set retrieval.

ToolRet-train is a separate training collection. Completeness requires every target tool to appear among the retrieved results.

Released
March 2025
Kind
Function calling
Level
University
Scored
Checked automatically
Answers
Short answer
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone

In the authors’ words

Tool learning aims to augment large language models (LLMs) with diverse tools, enabling them to act as agents for solving practical tasks. Due to the limited context length of tool-using LLMs, adopting information retrieval (IR) models to select useful tools from large toolsets is a critical initial step. However, the performance of IR models in tool retrieval tasks remains underexplored and unclear. Most tool-use benchmarks simplify this step by manually pre-annotating a small set of relevant tools for each task, which is far from the real-world scenarios. In this paper, we propose ToolRet, a heterogeneous tool retrieval benchmark comprising 7.6k diverse retrieval tasks, and a corpus of 43k tools, collected from existing datasets. We benchmark six types of models on ToolRet. Surprisingly, even the models with strong performance in conventional IR benchmarks, exhibit poor performance on ToolRet. This low retrieval quality degrades the task pass rate of tool-use LLMs. As a further step, we contribute a large-scale training dataset with over 200k instances, which substantially optimizes the tool retrieval ability of IR models.

Abstract of Retrieval Models Aren't Tool-Savvy: Benchmarking Tool Retrieval for Large Language Models (arXiv:2503.01763, CC0)
23
Entries in its paper, 0 models we track

bge-reranker-v2-gemma tops the leaderboard with 54.8

Best score per entry from the benchmark's paper, read on November 18th, 2026. A paper’s table doesn’t add newer models. 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.

  1. 1 bge-reranker-v2-gemma query only 54.8
  2. 2 GritLM-7B query only 51.8
  3. 3 bge-reranker-v2-m3 query only 49.8
  4. 4 NV-Embed-v1 query only 48.1
  5. 5 gte-Qwen2-1.5B-inst. query only 47.4
  6. 6 e5-mistral-7b query only 46.3
  7. 7 gte-base-en-v1.5 query only 44.0
  8. 8 bge-large-en-v1.5 query only 43.8
  9. 9 bm25 query only 43.1
  10. 10 bge-base-en-v1.5 query only 42.0
  11. 11 jina-reranker-v2-base query only 41.9
  12. 12 gtr-t5-large query only 40.9
  13. 13 gte-large-en-v1.5 query only 40.1
  14. 14 e5-large-v2 query only 39.5
  15. 15 e5-base-v2 query only 39.0
  16. 16 e5-small-v2 query only 38.3
  17. 17 gtr-t5-base query only 37.0
  18. 18 monot5-base-msmarco query only 35.7
  19. 19 contriever-msmarco query only 35.1
  20. 20 COLT query only 34.9
  21. 21 all-MiniLM-L6-v2 query only 32.0
  22. 22 mxbai-rerank-large-v1 query only 31.3
  23. 23 Colbert query only 30.5

Source: The ToolRet paper

#EntryScoreSettingDateSource
1 bge-reranker-v2-gemma 54.8 query only Official leaderboard
2 GritLM-7B 51.8 query only Official leaderboard
3 bge-reranker-v2-m3 49.8 query only Official leaderboard
4 NV-Embed-v1 48.1 query only Official leaderboard
5 gte-Qwen2-1.5B-inst. 47.4 query only Official leaderboard
6 e5-mistral-7b 46.3 query only Official leaderboard
7 gte-base-en-v1.5 44.0 query only Official leaderboard
8 bge-large-en-v1.5 43.8 query only Official leaderboard
9 bm25 43.1 query only Official leaderboard
10 bge-base-en-v1.5 42.0 query only Official leaderboard
11 jina-reranker-v2-base 41.9 query only Official leaderboard
12 gtr-t5-large 40.9 query only Official leaderboard
13 gte-large-en-v1.5 40.1 query only Official leaderboard
14 e5-large-v2 39.5 query only Official leaderboard
15 e5-base-v2 39.0 query only Official leaderboard
16 e5-small-v2 38.3 query only Official leaderboard
17 gtr-t5-base 37.0 query only Official leaderboard
18 monot5-base-msmarco 35.7 query only Official leaderboard
19 contriever-msmarco 35.1 query only Official leaderboard
20 COLT 34.9 query only Official leaderboard
21 all-MiniLM-L6-v2 32.0 query only Official leaderboard
22 mxbai-rerank-large-v1 31.3 query only Official leaderboard
23 Colbert 30.5 query only Official 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

All 8 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
  • Pass Independent runs Scores from the benchmark’s paper (23 entries).
  • Pass Models scored 23 entries on its leaderboard, none a model we track.
  • Note Newest score No scores yet.
  • Note Frontier models No models scored yet.
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 23 entries at the last read, none a model we track.
  • Pass Last read November 18th, 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