GPT-5 Unsupported parameter reasoning.effort

I can't use GPT-5
Cleaning .cache/opencode doesn't help
AI_APICallError: Unsupported parameter: 'reasoning.effort' is not supported with this model.

opencode v0.4.26
API provider: OpenAI

Same issue for me, but only when using "opencode run"

this working for me - what is the version here cat ~/.cache/opencode/node_modules/@ai-sdk/openai/package.json

what is the version

cat ~/.cache/opencode/node_modules/@ai-sdk/openai/package.json

{
  "name": "@ai-sdk/openai",
  "version": "2.0.10",
  "license": "Apache-2.0",
  "sideEffects": false,
  "main": "./dist/index.js",
  "module": "./dist/index.mjs",
  "types": "./dist/index.d.ts",
  "files": [
    "dist/**/*",
    "CHANGELOG.md",
    "internal.d.ts"
  ],
  "exports": {
    "./package.json": "./package.json",
    ".": {
      "types": "./dist/index.d.ts",
      "import": "./dist/index.mjs",
      "require": "./dist/index.js"
    },
    "./internal": {
      "types": "./dist/internal/index.d.ts",
      "import": "./dist/internal/index.mjs",
      "module": "./dist/internal/index.mjs",
      "require": "./dist/internal/index.js"
    }
  },
  "dependencies": {
    "@ai-sdk/provider": "2.0.0",
    "@ai-sdk/provider-utils": "3.0.1"
  },
  "devDependencies": {
    "@types/node": "20.17.24",
    "tsup": "^8",
    "typescript": "5.8.3",
    "zod": "3.25.76",
    "@vercel/ai-tsconfig": "0.0.0"
  },
  "peerDependencies": {
    "zod": "^3.25.76 || ^4"
  },
  "engines": {
    "node": ">=18"
  },
  "publishConfig": {
    "access": "public"
  },
  "homepage": "https://ai-sdk.dev/docs",
  "repository": {
    "type": "git",
    "url": "git+https://github.com/vercel/ai.git"
  },
  "bugs": {
    "url": "https://github.com/vercel/ai/issues"
  },
  "keywords": [
    "ai"
  ],
  "scripts": {
    "build": "pnpm clean && tsup --tsconfig tsconfig.build.json",
    "build:watch": "pnpm clean && tsup --watch",
    "clean": "rm -rf dist *.tsbuildinfo",
    "lint": "eslint \"./**/*.ts*\"",
    "type-check": "tsc --build",
    "prettier-check": "prettier --check \"./**/*.ts*\"",
    "test": "pnpm test:node && pnpm test:edge",
    "test:update": "pnpm test:node -u",
    "test:watch": "vitest --config vitest.node.config.js",
    "test:edge": "vitest --config vitest.edge.config.js --run",
    "test:node": "vitest --config vitest.node.config.js --run"
  }
}

do you have any additional config?

do you have any additional config?

No, fresh installation.

Pick the gpt-5 option instead of gpt-5-latest. It worked for me

made PR, should fix

on 0.4.37, and private openai azure endpoint:

I see error with 'reasoningEffort'. Did you mean 'reasoning_effort'
I see "text_verbosity":"low" being sent 2x ( I do have it this way 'text_verbosity' specified in subagent ), it's not unified on requestBodyValues.

Partial log:

ERROR 2025-08-13T05:51:31 +5643ms service=session session=ses_75e04927cffennnv5Mz3Erx5Jf error={"error":{"name":"AI_APICallError","url":"......
....
"requestBodyValues":{"model":"gpt-5","max_tokens":32000,"reasoningEffort":"minimal","textVerbosity":"low","name":"myTravelAgent","reasoning_effort":"minimal","text_verbosity":"low","messages":[{"role":"system","content":"You are an AI .......
...
,"responseBody":"{\"detail\":\"litellm.BadRequestError: AzureException BadRequestError - Unknown parameter: 'reasoningEffort'. Did you mean 'reasoning_effort'?. Received Model Group=gpt-5\\nAvailable Model Group Fallbacks=None\"}","isRetryable":false}
INFO  2025-08-13T05:51:31 +1ms service=bus type=session.error publishing
ERROR 2025-08-13T05:51:31 +3ms service=tui message=AI_APICallError: Bad Request name=UnknownError Server error
INFO  2025-08-13T05:51:31 +0ms service=bus type=storage.write publishing
INFO  2025-08-13T05:51:31 +0ms service=bus type=message.updated publishing

made PR, should fix

v0.0.41 still AI_APICallError: Unsupported parameter: 'reasoning.effort' is not supported with this model.

on private azure instance, using opencode v0.5.13 I still have:

  "provider": {
    "XXXX": {
      "npm": "@ai-sdk/openai-compatible",     // tested with this or even without
      "name": "YYYY",
      "options": {
        "baseURL": "https://ai.yyyy.com/openai/",
        "apiKey": "{env:OPENAI_API_KEY}"
      },
      "models": {
        "gpt-5": { 
          "name": "gpt-5"
        },

opencode run "hi" --model "XXXX/gpt-5"
Error: AI_APICallError: litellm.BadRequestError: AzureException BadRequestError - Unknown parameter: 'reasoningEffort'. Did you mean 'reasoning_effort'?. Received Model Group=gpt-5
Available Model Group Fallbacks=None

when provider XXXX is replaced with "openai" and npm is commented then it works without problems. After all looks like new users wont have a problem, I went to this as I have had previous 4.1 configuration in place.

on private azure instance, using opencode v0.5.13 I still have:
``` "provider": { "XXXX": { "npm": "@ai-sdk/openai-compatible", // tested with this or even without "name": "YYYY", "options": { "baseURL": "https://ai.yyyy.com/openai/", "apiKey": "{env:OPENAI_API_KEY}" }, "models": { "gpt-5": { "name": "gpt-5" }, ```
opencode run "hi" --model "XXXX/gpt-5"
Error: AI_APICallError: litellm.BadRequestError: AzureException BadRequestError - Unknown parameter: 'reasoningEffort'. Did you mean 'reasoning_effort'?. Received Model Group=gpt-5
Available Model Group Fallbacks=None

when provider XXXX is replaced with "openai" and npm is commented then it works without problems. After all looks like new users wont have a problem, I went to this as I have had previous 4.1 configuration in place.

But then I really have to call it opencode run "hi" --model "openai/gpt-5" with opencode run "hi" --model "YYYY/gpt-5" it will again fail on

ERROR 2025-08-22T11:23:10 +0ms service=tui message=AI_APICallError: litellm.BadRequestError: AzureException BadRequestError - Unknown parameter: 'reasoningEffort'. Did you mean 'reasoning_effort'?. Received Model Group=gpt-5

Which is a regression of previous error. Related to provider not having in name "openai".

Doing this also works

  "provider": {
    "CUSTOM_NAME": {
      "npm": "@ai-sdk/openai",     // tested with this or even without
      "name": "YYYY",
      "options": {
        "baseURL": "https://ai.yyyy.com/openai/",
        "apiKey": "{env:OPENAI_API_KEY}"
      },
      "models": {
        "gpt-5": { 
          "name": "gpt-5"
        },

I'm talking to an OpenAI compatible API (LiteLLM) and gpt-5 and gpt-5-mini work with out an issue.

I hope this helps, lets me keep things organized too.

Navigation Menu

Repository navigation

GPT-5 Unsupported parameter reasoning.effort #1859

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

GPT-5 Unsupported parameter reasoning.effort #1859

Description

Activity

Askath commented on Aug 12, 2025

thdxr commented on Aug 12, 2025

Salmin commented on Aug 12, 2025

thdxr commented on Aug 12, 2025

Salmin commented on Aug 12, 2025

vitorfdl commented on Aug 13, 2025

rekram1-node commented on Aug 13, 2025

f5epcim commented on Aug 13, 2025

Salmin commented on Aug 13, 2025

f5epcim commented on Aug 22, 2025

epcim commented on Aug 22, 2025

ncecere commented on Aug 24, 2025

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions