BFCL
Also known as Berkeley Function Calling Leaderboard, BFCL V4
BFCL, the Berkeley Function Calling Leaderboard, measures whether a model can call functions and APIs correctly. That means selecting the right function, filling its parameters with valid types and values, and declining to invent a function that was never offered. It is the number to read before wiring a model into your own tools, and it is conspicuously absent from most launch slides.
| What it measures | Whether a model calls functions and APIs correctly: picking the right function, filling parameters with valid types, and refusing to invent functions that were not offered. |
|---|---|
| Built by | UC Berkeley Gorilla team, 2024 |
| Format | Simple, multiple, parallel and nested function calls plus multi-turn interactions; V4 (2026) extends to holistic agentic evaluation including web search |
| Scoring metric | Abstract-syntax-tree match against a reference call, plus executable checks; overall score is the unweighted mean of subcategories |
| Status | Active |
| Representative top score | Not independently confirmed; see the leaderboard |
| Official leaderboard | gorilla.cs.berkeley.edu/leaderboard.html |
How BFCL works
The benchmark covers simple, multiple, parallel and nested function calls, plus multi-turn interactions. Scoring is primarily done with a deterministic abstract-syntax-tree matcher that compares the model’s invocation against a reference call, alongside executable checks where a real call can be run. V4, released in 2026, extends the suite toward holistic agentic evaluation including web search. The overall score is the unweighted mean of the subcategories.
History and current status
The UC Berkeley Gorilla team introduced BFCL in 2024, and it became the default reference for tool-calling reliability as function calling turned into the foundation of every agent product. Successive versions added abstract-syntax-tree metrics, enterprise and open-source function sets, and multi-turn interaction. V4 in 2026 moved from testing calls in isolation to evaluating agentic behaviour end to end.
What the score does not tell you
Matching against a reference invocation rewards format discipline as much as judgment: a model that makes a sensible but differently structured call can be marked wrong. The unweighted mean across subcategories also means a model can post a respectable overall score while being weak in the one category you depend on, so the subcategory breakdown matters far more than the headline number.
This is the general case, not a quirk of one benchmark. For the four failure modes that decide whether any benchmark number means anything, read are AI benchmarks reliable, and for this row in particular, how benchmark contamination inflates scores.
Who reports BFCL, and how to read it
Berkeley runs the leaderboard, which makes it academically independent, and it is updated continuously. Frontier labs rarely lead with BFCL, so if you need to know whether a model will respect your tool schemas, you generally have to go and read the board rather than the launch post.
Benchmarks to read alongside this one
tau-bench
Whether a tool-using agent can reliably complete customer-service tasks over multi-turn conversations with a simulated user while obeying domain policies.
GAIA
Whether an AI assistant can answer real-world questions that require multi-step reasoning, multiple modalities, web browsing and general tool use.
IFEval
Whether a model obeys instructions that can be checked by a program, such as a word count, a required keyword, or a forbidden format.
BFCL: frequently asked questions
- What is BFCL?
- BFCL is the Berkeley Function Calling Leaderboard, an academic benchmark measuring whether a model calls functions and APIs correctly: picking the right function, supplying valid parameter types and values, and not hallucinating functions that were not provided.
- How is BFCL scored?
- Mainly with a deterministic abstract-syntax-tree matcher that compares the predicted call against a reference invocation, plus executable checks where a call can actually be run. The overall figure is the unweighted average across subcategories such as simple, multiple, parallel and nested calls.
- What is new in BFCL V4?
- V4, released in 2026, shifts from evaluating function calls in isolation to holistic agentic evaluation, including web search. Earlier versions had added the abstract-syntax-tree metric, enterprise and open-source function sets, and multi-turn interaction.
- Why does function-calling accuracy matter more than benchmark scores?
- Because in production a model that reasons well but formats a tool call wrongly simply fails. Function calling is the interface between the model and everything it can actually do, so its reliability caps the reliability of the whole system regardless of reasoning ability.
Get each breakdown before it makes the rounds
You get one email when a new source-backed analysis goes live: what AI agents actually cost, which models are worth running, and what the benchmarks really mean. No hype.
No spam. Unsubscribe anytime.