Back to changelog

Dev Plans, Native Web Search, and MiniMax Provider

Ship faster with Dev Plans — AI-powered development planning now in beta. Plus native web search for real-time data, MiniMax provider, structured outputs for Anthropic & Perplexity, and a redesigned models experience.

Dev Plans dashboard and web search capabilities

Dev Plans: AI-Powered Development Planning (Beta)

We're launching Dev Plans — a new way to plan and execute software projects with AI assistance. Break down complex features into actionable steps, get implementation guidance, and ship faster.

Try Dev Plans now — we're looking for early feedback to shape the product.

What you can do

  • Plan features — describe what you want to build and get a structured implementation plan
  • Break down tasks — complex projects split into manageable, actionable steps
  • Get code guidance — AI-assisted implementation recommendations

Dev Plans is available on all paid plans. We'd love your feedback as we iterate on this feature.

Native Web Search

LLM responses can now include real-time web data. No more outdated information — your AI assistant can search the web to answer questions about current events, recent releases, or live data.

How it works

Web search is billed per search call and works seamlessly with supported models. Read the docs

1curl -X POST https://api.llmgateway.io/v1/chat/completions \
2  -H "Authorization: Bearer $LLM_GATEWAY_API_KEY" \
3  -H "Content-Type: application/json" \
4  -d '{
5    "model": "openai/gpt-4o",
6    "messages": [{"role": "user", "content": "What are the latest AI announcements this week?"}],
7    "web_search": true
8  }'

MiniMax Provider

We've added MiniMax as a new provider, expanding your options for high-quality language models. MiniMax offers competitive pricing and strong performance across various tasks.

View MiniMax models

Structured Outputs for Anthropic & Perplexity

Get reliable JSON responses with structured outputs — now available for Anthropic Claude models and Perplexity. Define your schema and get guaranteed valid JSON back.

1curl -X POST https://api.llmgateway.io/v1/chat/completions \
2  -H "Authorization: Bearer $LLM_GATEWAY_API_KEY" \
3  -H "Content-Type: application/json" \
4  -d '{
5    "model": "anthropic/claude-sonnet-4-20250514",
6    "messages": [{"role": "user", "content": "Extract the name and email from: John Smith, john@example.com"}],
7    "response_format": {
8      "type": "json_schema",
9      "json_schema": {
10        "name": "contact",
11        "schema": {
12          "type": "object",
13          "properties": {
14            "name": {"type": "string"},
15            "email": {"type": "string"}
16          }
17        }
18      }
19    }
20  }'

Redesigned Models Experience

The models page got a major upgrade:

  • Flattened table structure — easier to scan and compare models at a glance
  • Clickable model IDs — copy model identifiers with one click
  • Web search pricing — see which models support web search and at what cost
  • Coding models filter — quickly find models optimized for code generation

Explore the new models page

New Models

GLM-4.7 Family

The latest GLM models are now available across multiple providers:

zai/glm-4.7-flash — fast and efficient

1zai/glm-4.7-flash

Try in Playground

zai/glm-4.7-flashx — extended context

1zai/glm-4.7-flashx

Try in Playground

cerebras/glm-4.7 — ultra-low latency via Cerebras

1cerebras/glm-4.7

Try in Playground

novita/glm-4.7-flash — cost-effective option

1novita/glm-4.7-flash

Try in Playground

Image Generation

zai/cogview-4 — advanced image generation from Z.AI

1zai/cogview-4

zai/glm-image — text-to-image with excellent text rendering

1zai/glm-image

bytedance/seedream-4.0 — high-quality image synthesis

1bytedance/seedream-4.0

bytedance/seedream-4.5 — high-quality image synthesis

1bytedance/seedream-4.5

ByteDance ModelArk

Eight new models from ByteDance including GPT-OSS-120B variants. View all ByteDance models

More Improvements

  • JSON Response Healing — malformed JSON from models is automatically repaired. Read the docs
  • API Key Usage Filters — filter your usage statistics by specific API keys. Read the docs
  • Playground Upgrades — AI SDK v6, Web Search, improved scrolling
  • Better Error Handling — improved fallback routing when providers have issues. Read the docs

Try Dev Plans — your feedback shapes the product.

Explore new models — find the right model for your use case.

Get started — free tier available.