Compare AI Models Side by Side

Pick 2 to 4 models to compare benchmark scores, pricing, and specs in 2026.

GLM-5.2
Tencent Hy3
Claude Sonnet 4.6
GPT-5.5
Z.ai
GLM-5.2

Z.ai · GLM

Tencent
Tencent Hy3

Tencent · Hy

Anthropic
Claude Sonnet 4.6

Anthropic · Claude

OpenAI
GPT-5.5

OpenAI · GPT

Specifications

Spec
GLM-5.2
Tencent Hy3
Claude Sonnet 4.6
GPT-5.5
ProviderZ.aiTencentAnthropicOpenAI
FamilyGLMHyClaudeGPT
Release date2026-06-162026-07-062026-02-172026-04-23
Open sourceYesYesNoNo
Context window1.0M tokens262K tokens1.0M tokens1.0M tokens
Max output131K tokens--128K tokens--
Input price / MTok$0.90*$0.20*$3.00$5.00
Output price / MTok$2.86*$0.80*$15.00$30.00
Speed----47 tok/s89 tok/s
Best forcoding, reasoning, agentscoding, reasoning, agentscoding, agents, visioncoding, reasoning, agents

* GLM-5.2: Via OpenRouter (model ID z-ai/glm-5.2); SiliconFlow lists $1.40/$4.40 per MTok | Tencent Hy3: Free via OpenRouter for ~2 weeks from 2026-07-07 (model ID tencent/hy3:free); paid tier $0.20/$0.80 per MTok via OpenRouter

Benchmark scores (% of max)

Scores normalized to percentage of each benchmark's maximum reported value. Only officially reported scores are included.

Cost per million tokens (USD)

Context window (thousands of tokens)

Output speed (tokens per second)

Notes

GLM-5.2: 744B total / 40B active MoE; 1M-token context with IndexShare attention (2.9x per-token FLOPs reduction at full context). MIT license. Two reasoning modes: High and Max.
Tencent Hy3: 295B total / 21B active MoE (192 experts, top-8 routing); no-think plus low/high chain-of-thought modes. Apache 2.0.
GPT-5.5: OpenAI's current flagship; GPT-5.5 Instant became ChatGPT's default for free users May 5 2026.
Back to all models

Related

Deciding which model to build on for a real product? Book a free consultation and we'll help you pick the right one for your workload and budget.

GLM-5.2 vs Tencent Hy3 vs Claude Sonnet 4.6 vs GPT-5.5 | AI Model Comparison