LLM Model Comparison
15 models across 6 providers — capabilities, pricing, and use cases · July 2026
Anthropic Claude (cc/)
Claude Opus 4.8
Anthropic · cc/claude-opus-4-8
Flagship reasoning model. Top-tier coding (SWE-bench leader), deep analysis, extended thinking.
Context
200K tokens
Tier
Flagship
Input
$15 / M
Output
$75 / M
Best For
Complex software engineering & architecture
Multi-step research & analysis
Agentic workflows & tool use
Claude Sonnet 5
Anthropic · cc/claude-sonnet-5
Balanced performance/cost. Strong coding, writing, and analysis. Production workhorse.
Context
200K tokens
Tier
Balanced
Input
$3 / M
Output
$15 / M
Best For
Production chatbots & assistants
Content generation & editing
Code review & moderate tasks
Claude Fable 5
Anthropic · cc/claude-fable-5
Creative-focused variant. Optimized for storytelling, roleplay, and imaginative content.
Context
200K tokens
Tier
Creative
Input
$3 / M
Output
$15 / M
Best For
Creative writing & fiction
Game narratives & character dev
Brainstorming & ideation
Claude Haiku 4.5
Anthropic · cc/claude-haiku-4-5
Fastest & cheapest Claude. Great for high-volume, low-latency tasks.
Context
200K tokens
Tier
Fast
Input
$0.80 / M
Output
$4 / M
Best For
High-volume classification & routing
Real-time chat responses
Data extraction & summarization
OpenAI GPT-5.6 (cx/)
GPT-5.6 Sol
OpenAI · cx/gpt-5.6-sol
OpenAI flagship. Highest capability across all domains — coding, reasoning, analysis, creativity.
Context
128K tokens
Tier
Flagship
Input
$10 / M
Output
$30 / M
Best For
Complex multi-domain tasks
Research & deep analysis
Mission-critical applications
GPT-5.6 Terra
OpenAI · cx/gpt-5.6-terra
Mid-tier balance. Good capability at moderate cost. Sweet spot for production workloads.
Context
128K tokens
Tier
Balanced
Input
$2.50 / M
Output
$10 / M
Best For
Production chatbots
Content generation at scale
Code assistance (moderate)
GPT-5.6 Luna
OpenAI · cx/gpt-5.6-luna
Fastest & cheapest GPT-5.6. Optimized for speed and cost.
Context
128K tokens
Tier
Fast
Input
$0.87 / M
Output
$3.50 / M
Best For
High-volume API calls
Simple Q&A & classification
Drafting & templated content
xAI Grok (xai/)
Grok 4.5
xAI · xai/grok-4.5
xAI flagship with real-time X/Twitter data access. Strong general reasoning with live data advantage.
Context
131K tokens
Tier
Flagship
Input
$3 / M
Output
$15 / M
Best For
Trending topics & social listening
Real-time news analysis
General assistant with live context
OCG Provider Models (ocg/)
GLM-5.2
Zhipu AI · ocg/glm-5.2
Open-weight MoE (~753B params). MIT licensed. 1M context. Best-in-class agentic coding.
Context
1M tokens
Tier
Flagship
Input
~$0.15 / M
Output
~$0.60 / M
Best For
Repository-level code generation
Long-horizon coding agents
Terminal-based automation
Kimi K2.7 Code
Moonshot AI · ocg/kimi-k2.7-code
Code-specialized. Optimized for software engineering, debugging, full-stack development.
Context
128K tokens
Tier
Code
Input
~$0.07 / M
Output
~$0.21 / M
Best For
Code generation & debugging
Code review & refactoring
Full-stack development
DeepSeek V4 Flash
DeepSeek · ocg/deepseek-v4-flash
Fast variant. Trades capability for speed & cost. Extremely cheap. Strong general capability.
Context
128K tokens
Tier
Fast
Input
~$0.03 / M
Output
~$0.10 / M
Best For
High-volume, cost-sensitive apps
Real-time chat & conversational AI
Batch processing & classification
MiMo V2.5
Xiaomi · ocg/mimo-v2.5
Reasoning-focused. Strong chain-of-thought, math, logic. Top Chinese model rankings. Open weights.
Context
128K tokens
Tier
Reasoning
Input
Free (open)
Output
Free (open)
Best For
Mathematical reasoning & problem solving
Logic-heavy tasks & step-by-step
Educational AI applications
MiniMax M3
MiniMax · ocg/minimax-m3
Large context model. Open weights. Strong general capability with 1M+ context window.
Context
1M+ tokens
Tier
Balanced
Input
~$0.10 / M
Output
~$0.50 / M
Best For
Long document analysis
Enterprise conversational AI
Multi-document research
Local / Self-Hosted (ollama/)
Gemma Heretic Q4_K_M
Google (community) · ollama/gemma-heretic:q4km
Community-modified Google Gemma, quantized for local deployment. Fully offline, zero API cost.
Context
8K-32K tokens
Tier
Local
Input
Free (self-hosted)
Output
Free (self-hosted)
Best For
Privacy-sensitive data processing
Offline coding assistance
Experimentation & learning
big-usage
Unknown · big-usage
Model not found in public registries. May be a custom/private endpoint or internal alias.
Context
Unknown
Tier
?
Input
Unknown
Output
Unknown
Status
Could not be found in any public registry
Verify source — may be internal alias
Side-by-Side Comparison
| Model | Provider | Tier | Context | Input $/M | Output $/M | Coding | Reasoning | Creative | Speed | Best Use |
|---|---|---|---|---|---|---|---|---|---|---|
| Claude Opus 4.8 | Anthropic | Flagship | 200K | $15.00 | $75.00 | ★★★★★ | ★★★★★ | ★★★★☆ | ★★★☆☆ | Complex engineering |
| Claude Sonnet 5 | Anthropic | Balanced | 200K | $3.00 | $15.00 | ★★★★☆ | ★★★★☆ | ★★★★☆ | ★★★★☆ | Production workhorse |
| Claude Fable 5 | Anthropic | Creative | 200K | $3.00 | $15.00 | ★★★☆☆ | ★★★☆☆ | ★★★★★ | ★★★★☆ | Creative writing |
| Claude Haiku 4.5 | Anthropic | Fast | 200K | $0.80 | $4.00 | ★★★☆☆ | ★★★☆☆ | ★★★☆☆ | ★★★★★ | High-volume tasks |
| GPT-5.6 Sol | OpenAI | Flagship | 128K | $10.00 | $30.00 | ★★★★★ | ★★★★★ | ★★★★★ | ★★★☆☆ | Mission-critical |
| GPT-5.6 Terra | OpenAI | Balanced | 128K | $2.50 | $10.00 | ★★★★☆ | ★★★★☆ | ★★★★☆ | ★★★★☆ | Production chatbots |
| GPT-5.6 Luna | OpenAI | Fast | 128K | $0.87 | $3.50 | ★★★☆☆ | ★★★☆☆ | ★★★☆☆ | ★★★★★ | Simple tasks, drafts |
| Grok 4.5 | xAI | Flagship | 131K | $3.00 | $15.00 | ★★★★☆ | ★★★★☆ | ★★★★☆ | ★★★★☆ | Real-time social data |
| GLM-5.2 | Zhipu AI | Flagship | 1M | ~$0.15 | ~$0.60 | ★★★★★ | ★★★★☆ | ★★★☆☆ | ★★★★☆ | Agentic coding |
| Kimi K2.7 Code | Moonshot | Code | 128K | ~$0.07 | ~$0.21 | ★★★★☆ | ★★★☆☆ | ★★☆☆☆ | ★★★★☆ | Code gen & debugging |
| DeepSeek V4 Flash | DeepSeek | Fast | 128K | ~$0.03 | ~$0.10 | ★★★★☆ | ★★★☆☆ | ★★★☆☆ | ★★★★★ | Ultra-cheap bulk |
| MiMo V2.5 | Xiaomi | Reasoning | 128K | Free | Free | ★★★☆☆ | ★★★★★ | ★★☆☆☆ | ★★★★☆ | Math, logic, education |
| MiniMax M3 | MiniMax | Balanced | 1M+ | ~$0.10 | ~$0.50 | ★★★☆☆ | ★★★☆☆ | ★★★☆☆ | ★★★☆☆ | Long doc analysis |
| Gemma Heretic | Local | 8-32K | Free | Free | ★★☆☆☆ | ★★☆☆☆ | ★★☆☆☆ | ★★★★★ | Privacy, offline | |
| big-usage | Unknown | ? | ? | ? | ? | ? | ? | ? | ? | Unverified |
Use Case → Model Recommendation
🖥️ Full Stack Web Development
Complex coding, architecture, multi-file refactoring, production debugging.
→ Claude Opus 4.8 · GLM-5.2 · Kimi K2.7 Code
📱 WhatsApp Bot / Chat
Real-time conversational AI, fast responses, cost-effective at scale.
→ Claude Haiku 4.5 · GPT-5.6 Luna · DeepSeek V4 Flash
💰 Receipt OCR / Data Extraction
Image understanding, structured extraction, classification.
→ Claude Sonnet 5 · GPT-5.6 Terra · MiMo V2.5
📊 Financial Analysis
Numerical reasoning, tax calculations, budget projections.
→ MiMo V2.5 · Claude Opus 4.8 · GPT-5.6 Sol
✍️ Content & Copywriting
Blog posts, social media, marketing copy, brand voice.
→ Claude Fable 5 · Claude Sonnet 5 · GPT-5.6 Terra
🔍 OSINT / Research
Data gathering, analysis, pattern recognition, reports.
→ Grok 4.5 · Claude Opus 4.8 · GLM-5.2
🎓 Education / Tutoring
Step-by-step explanations, math, concept teaching.
→ MiMo V2.5 · Claude Sonnet 5 · Gemma Heretic
📄 Document Processing
Long document summarization, contract review, analysis.
→ MiniMax M3 · GLM-5.2 · Claude Sonnet 5
🔒 Privacy-First / Offline
No data leaves machine, sensitive data, air-gapped.
→ Gemma Heretic Q4_K_M (only option)
⚡ High-Volume Batch
Classification, routing, tagging at scale.
→ DeepSeek V4 Flash · GPT-5.6 Luna · Claude Haiku 4.5
💰 Best Budget Pick
Maximum capability per dollar. Free tiers & open weights.
→ MiMo V2.5 (free) · DeepSeek V4 Flash ($0.03/M)
🏆 Best Overall
Highest capability regardless of cost.
→ Claude Opus 4.8 or GPT-5.6 Sol
Cost: 1M Input + 200K Output Tokens
| Model | Input | Output | Total | vs Cheapest |
|---|---|---|---|---|
| MiMo V2.5 | $0.00 | $0.00 | $0.00 | — |
| Gemma Heretic | $0.00 | $0.00 | $0.00 | — |
| DeepSeek V4 Flash | $0.03 | $0.02 | $0.05 | 1x |
| Kimi K2.7 Code | $0.07 | $0.04 | $0.11 | 2x |
| MiniMax M3 | $0.10 | $0.10 | $0.20 | 4x |
| GLM-5.2 | $0.15 | $0.12 | $0.27 | 5x |
| Claude Haiku 4.5 | $0.80 | $0.80 | $1.60 | 32x |
| GPT-5.6 Luna | $0.87 | $0.70 | $1.57 | 31x |
| GPT-5.6 Terra | $2.50 | $2.00 | $4.50 | 90x |
| Claude Sonnet 5 | $3.00 | $3.00 | $6.00 | 120x |
| Grok 4.5 | $3.00 | $3.00 | $6.00 | 120x |
| GPT-5.6 Sol | $10.00 | $6.00 | $16.00 | 320x |
| Claude Opus 4.8 | $15.00 | $15.00 | $30.00 | 600x |