AI models with tool calling: 166 that can call your functions

Prices from official vendor price lists · October 2026
166 models that call your functions and external APIs with structured arguments — the foundation of any integration. Output prices run from $0.059 to $180 per 1M tokens across the 142 models with a paid per-token price; 58 have a free tier and 76 ship open weights.
Tool calling
Services found: 166
#AI modelCapabilitiesAPI $/1M (in / out)Context windowPrice / moOutput $/1MOverviewVisit site
41

Grok 4.3

xAI
Text & chatVisionReasoning
$1.25 / $2.51M tokensAPI only2.5View
42

DeepSeek V4 Flash 0423

DeepSeek
Text & chatVisionReasoning
$0.088606 / $0.1772121.05M tokensAPI only0.177View
43

GPT-5.4 Mini

OpenAI
Text & chatVisionReasoning
$0.75 / $4.5400K tokensAPI only4.5View
44

Nemotron 3 Ultra

NVIDIA
Text & chatReasoningTool calling
$0.625 / $3.125262.1K tokensAPI only3.125View
45

MiniMax M2.7

MiniMax
Text & chatReasoningTool calling
$0.3 / $1.2204.8K tokensAPI only1.2View
46

Gemini 3.5 Flash-Lite

Google
Text & chatVisionAudio
$0.3 / $2.51.05M tokensAPI only2.5View
47

Qwen3.6 27B

Qwen
Text & chatVisionVideo
$0.3 / $2262.1K tokensAPI only2View
48

GPT-5.4 Nano

OpenAI
Text & chatVisionReasoning
$0.2 / $1.25400K tokensAPI only1.25View
49

Claude Sonnet 4.5

Anthropic
Text & chatVisionReasoning
$3 / $151M tokensAPI only15View
50

Qwen3.5 397B A17B

Qwen
Text & chatVisionVideo
$0.55 / $3.5262.1K tokensAPI only3.5View
51

Claude Haiku 4.5

Anthropic
Text & chatVisionReasoning
$1 / $5200K tokensAPI only5View
52

Qwen3.5-122B-A10B

Qwen
Text & chatVisionVideo
$0.29 / $2.4262.1K tokensAPI only2.4View
53

Mistral Medium 3.5

Mistral AI
Text & chatVisionReasoning
$1.5 / $7.5256K tokensAPI only7.5View
54

Nemotron 3.5 Lightning

NVIDIA
Text & chatReasoningTool calling
$0.08 / $0.2262.1K tokensAPI only0.2View
55

Nemotron 3 Super

NVIDIA
Text & chatReasoningTool calling
$0.085 / $0.4262.1K tokensAPI only0.4View
56

Mistral Small 4

Mistral AI
Text & chatVisionReasoning
$0.15 / $0.6256K tokensAPI only0.6View
57

Trinity Large Thinking

Arcee Ai
Text & chatReasoningTool calling
$0.25 / $0.8262.1K tokensAPI only0.8View
58

Qwen3 Coder Next

Qwen
Text & chatCodingReasoning
$0.12 / $0.8262.1K tokensAPI only0.8View
59

Mistral Large 3 2512

Mistral AI
Text & chatVisionReasoning
$0.5 / $1.5262.1K tokensAPI only1.5View
60

Nemotron 3 Nano 30B A3B

NVIDIA
Text & chatReasoningTool calling
$0.05 / $0.2262.1K tokensAPI only0.2View

Browse by need

Reasoning models169 modelsCoding assistants17 modelsVision and images156 modelsLong context 1M+82 modelsFree tier available80 modelsOpen weights88 models

Tool calling is what turns a chatbot into software. The model returns a structured call - function name plus typed arguments - and your code executes it, feeds back the result and lets the model continue. Every integration worth building rests on it.

All 166 models here support it, from Claude Fable 5.1 at $50 per million output tokens down to budget options at $0.25. What varies is reliability under pressure: how well the model picks the right tool from a long list, and how many calls it can chain before losing the thread. Frontier models are noticeably better at long chains.

Structured, not guessedThe model returns a typed call, so your code never parses prose to decide what to run.
Chains break at scaleReliability drops as the tool catalogue grows; frontier models hold longer chains together.
Pairs with JSON outputTool calling picks the action, structured output guarantees the shape of the final answer.

Models with this capability

The 166 models below all support this capability. Prices are input and output per 1M tokens from the vendor official price list; for dirhams multiply by 3.6725. Tap a card to open the model page.

Claude Fable 5.1
#1
Anthropic
in $10out $501M tokens
GPT-6 Astra
#2
OpenAI
in $10out $501.05M tokens
GPT-6.1 Sol
#3
OpenAI
in $2out $101.05M tokens
GPT-6 Sol
#4
OpenAI
in $2out $101.05M tokens
GPT-6 Luna
#5
OpenAI
in $0.1out $0.51.05M tokens
Claude Opus 5.5
#6
Anthropic
in $4out $201M tokens
Claude Sonnet 5.5
#7
Anthropic
in $2out $101M tokens
Grok 4.7
#8
xAI
in $2out $6500K tokens

How to choose

  • Keep tool descriptions short and unambiguous — most failures are description problems, not model problems.
  • Limit the catalogue per request; twenty tools is manageable, two hundred is not.
  • Handle partial failure explicitly: models recover better when the error message says what to do next.
  • Cache the tool definitions — they are the same on every call and caching cuts that input cost.

Common mistakes

One giant toolSplitting into several narrow tools raises accuracy far more than prompt tuning.
No timeout on chainsWithout an iteration cap an agentic loop can call tools until the budget is gone.

Who this suits

Integration teamsConnecting an assistant to internal APIs, CRMs and databases.
AutomationWorkflows where the model decides the next step rather than following a script.
Support productsLooking up orders, issuing refunds, checking availability inside the conversation.

Related filters

Agentic workflows5 models built for autonomous multi-step work
Structured JSON151 models with schema-guaranteed output
Code execution47 models that run code in a sandbox
Reasoning169 models that think before answering
All 239 modelsThe full rating with filters and prices

Frequently asked questions

What is the difference from structured output?

Tool calling decides which function to run and with what arguments. Structured output constrains the shape of the final answer. Most production systems use both.

Do cheap models call tools reliably?

For a handful of well-described tools, yes. Accuracy drops as the catalogue grows and as chains get longer, which is where frontier models pull ahead.

Can the model run the tool itself?

Not by default — your code executes the call. For models that run code in a vendor sandbox, see code execution.

Can the model call several tools at once?

Most current models support parallel calls, which cuts latency on independent lookups.

What happens if the model invents a tool?

A well-formed implementation rejects unknown names and returns an error the model can act on.

Frequently searched

By capabilityReasoningCodingVisionLong context 1M+Free tierOpen weightsLow latencyRealtime voiceImage generation
By vendorAnthropic (13)OpenAI (30)Google DeepMind (17)xAI (8)Meta (6)DeepSeek (9)Mistral AI (11)Alibaba (Qwen) (34)Z.ai (12)Moonshot AI (5)
Popular modelsClaude Fable 5.1GPT-6 AstraClaude Opus 5Claude Fable 5GPT-5.6 SolGLM 5.3Grok 4.6Kimi K3GPT-5.6 Terra
By deliveryBatch APIPrompt cachingPriority tierFrontier tierProprietary APITool callingStructured JSONWeb searchMultilingual

Ratings 0

Avg. score 0

Rate this page

Help us improve