GPT-5.5
GPT-5.5 is worth evaluating for coding, rag, and agents when its provider route and context window match the workload.
Use it for
- Teams evaluating coding, rag, and agents
- Workloads that can use a 1.05m context window
- Buyers comparing 3 tracked provider routes
Do not use it for
- Workloads where another current model has stronger sourced task evidence
- Family
- GPT-5.5
- Released
- 2026-04-23
- Context
- 1.05m
- Max output
- 128,000
- Architecture
- Decoder Only
- Knowledge cutoff
- 2025-12
- Specialization
- general
- Openness
- Proprietary
- License
- ProprietaryCommercial use with conditions
Cheapest of 3 routes · OpenAI API · cache read $0.500
About
GPT-5.5 is OpenAI's fully retrained agentic model, released April 23, 2026. Optimised for agentic coding, computer use, knowledge work, and early scientific research. Achieves 82.7% on Terminal-Bench 2.0 (Codex CLI scaffold), 84.9% on GDPval, 58.6% on SWE-Bench Pro, 93.6% on GPQA Diamond, and 82.6% on SWE-Bench Verified (Vals.ai independent harness). Knowledge cutoff December 2025. Supports reasoning effort levels (none/low/medium/high/xhigh). Context window 1,050,000 tokens with a long-context surcharge above 272K tokens. Model ID: gpt-5.5.
GPT-5.5 is a proprietary model. The structured metadata tracks a 1.05m-token context window, multimodal input, reasoning, function calling, tool use, structured outputs, and code execution. This page tracks provider routes through OpenAI API, OpenRouter, and Vercel AI Gateway, with the cheapest tracked route listed at $5 input and $30 output per 1M tokens. Headline tracked benchmarks include Google-Proof Q&A 93.6, SWE-bench Pro 58.6, and Chatbot Arena 1488.0.
Top use-case fit: coding, agents, and build tasks
Coding
Q/$ D4 relevant benchmarks in the decision map.
RAG
Included by capability and metadata signals in the decision map.
Agents
Q/$ D1 relevant benchmark in the decision map.
Provider price ladder
Compare all 3Compare API pricing across 3 providers for input and output tokens, batch, and cached reads when available.
| Provider | Input / 1M | Output / 1M | Batch in / out | Cache | Route |
|---|---|---|---|---|---|
| OpenAI API | $5.00 | $30.00 | $2.50 / $15.00 | read $0.500 | Serverless |
| OpenRouter | $5.00 | $30.00 | - | - | Serverless |
| Vercel AI Gateway | $5.00 | $30.00 | - | read $0.500 | Serverless |
Available via routers & gateways(15)
AIRouter
RouterCommercial LLM router that analyzes incoming requests and routes to the optimal model for cost/quality/latency via a drop-in OpenAI-compatible API, with a privacy-preserving embedding mode that avoids sending prompt content.
Helicone
GatewayObservability-first AI gateway with routing, caching, rate limiting, and request tracing; Apache 2.0 open-source core with a managed hosted tier for logging and analytics.
Kong AI Gateway
GatewayMulti-LLM AI gateway built on Kong Gateway 3.x, adding semantic routing, load balancing, guardrails, and MCP traffic analytics as plugins over Kong's existing API management platform.
LiteLLM
GatewayOpen-source Python SDK and proxy server that unifies 100+ LLM APIs behind a single OpenAI-compatible interface, with load balancing, cost tracking, and configurable failover.
Martian
RouterAI-powered LLM router that analyzes each prompt in real-time to select the optimal model, targeting 20–97% cost reduction while maintaining quality; San Francisco startup reportedly nearing $1.3B valuation.
Neutrino AI
RouterCommercial LLM router that dynamically routes each query to the best-suited model with load balancing and fallback handling, charging 3% of underlying AI spend.
Capabilities
Benchmark peer barsfor Coding
Benchmark scores(17)
| Benchmark | Score | Version | Source |
|---|---|---|---|
| Google-Proof Q&A | 93.6 | diamond | https://openai.com/index/introducing-gpt-5-5/ |
| SWE-bench Pro | 58.6 | SWE-bench Pro (pass@1) | https://www.vellum.ai/blog/everything-you-need-to-know-about-gpt-5-5 |
| Chatbot Arena | 1488.0 | High | https://arena.ai/leaderboard |
| MMLU PRO | 88.1 | — | https://openai.com/index/introducing-gpt-5-5/ |
| MMMU Pro | 88.3 | Vals.ai standardized CoT harness | https://www.vals.ai/benchmarks/mmmu |
| HumanEval | 94.2 | — | https://openai.com/index/introducing-gpt-5-5/ |
| Instruction-Following Evaluation | 92.1 | — | https://openai.com/index/introducing-gpt-5-5/ |
| SWE-bench Verified | 82.6 | SWE-bench Verified | https://www.vals.ai/benchmarks/swebench |
| Massive Multitask Language Understanding | 92.4 | MMLU (accuracy) | https://tokenmix.ai/blog/gpt-5-5-spud-review-88-swe-bench-2026 |
| Terminal-Bench 2.0 | 82.7 | Terminal-Bench 2.0 (accuracy%) | https://llm-stats.com/benchmarks/terminal-bench-2 |
| Aider Polyglot | 88.0 | Listed as 'gpt-5 (high)' = 88 (percent_correct) | https://aider.chat/docs/leaderboards/ |
| AIME 2025 | 81.2 | AIME 2025 (accuracy) | https://codersera.com/blog/openai-may-2026-updates-roundup/ |
| ARC-AGI-2 | 85.0 | ARC-AGI-2 (accuracy%) | https://benchlm.ai/benchmarks/arcAgi2 |
| BrowseComp | 84.4 | BrowseComp (accuracy%) | https://benchlm.ai/benchmarks/browseComp |
| Humanity's Last Exam | 41.4 | HLE without tools (accuracy) | https://www.vellum.ai/blog/everything-you-need-to-know-about-gpt-5-5 |
| MCP-Atlas | 75.3 | MCP-Atlas (accuracy%) | https://benchlm.ai/benchmarks/mcpAtlas |
| GDPval | 84.9 | GDPval official launch score | https://openai.com/index/introducing-gpt-5-5/ |
Migration checks
Rankings & picks(10)
Comparison and alternatives
Browse all comparisons →Show all 72 popular comparisonssorted by 7-day search impressions
Cheapest of 3 routes · OpenAI API · cache read $0.500