Claude Opus 4.8
Version 4.8 of Anthropic's Claude Opus model series, capable of driving smartphone operations through the Android Remote Control MCP.
Timeline (last 90 days)
Z.ai claims that GLM-5.3-Flash is close to Claude Opus 4.8 on coding and agent benchmarks and costs one tenth of GLM-5.2's price.
The study used two frontier models, Claude Opus 4.8 and Codex GPT Sol 5.6, to reconstruct the Protobuf schema between the iPhone and Apple's location services directly from Apple's binary files.
Starting November 16, 2026, three models—Claude Opus 4.8, Claude Opus 5, and GPT-5.6 Terra—will be removed from GitHub Copilot's Chat, inline editing, ask and agent modes, and code completion.
Claude Opus 4.8 will continue to be available to individual Copilot users on annual subscription plans; for request-based annual Pro and Pro+ subscribers, Opus 4.8 has a multiplier of 27 (each request consumes 27 premium requests).
On a contamination-free benchmark, the trained 8B model surpassed Claude Opus 4.8 and GPT-6 Astra on the creative utility metric.
When a coding agent performed proof-of-concept experiments on these ideas, the model's success@k was comparable to that of Claude Opus 4.8.
In a preliminary human study involving four experts, the idea sets generated by this model were rated as equally promising as those of Claude Opus 4.8, and slightly more numerous in unique ideas.
Opus 5 and Opus 4.8 Fast Mode are still in research preview, and their requests still require the beta header.
Tev-27B achieves Claude Opus 4.8-level performance on the medium-horizon task TB 2.1.
The visual quality scores of AesCode-8B and AesCode-32B are comparable to closed-source frontier models such as GPT-5.5 and Claude Opus 4.8.
22 more timeline events
All Facts (20)
GLM-5.3-Flash achieved a score of 57 on the Artificial Analysis composite intelligence index evaluation, on par with Claude Opus 4.8.
70%UnverifiedA benchmark test conducted by Databricks on their own codebase of one million lines showed that Pi outperformed Claude Code and Codex in most scenarios, with the Pi + Claude Opus combination achieving the highest code quality.
60%UnverifiedZ.ai claims that GLM-5.3-Flash is close to Claude Opus 4.8 on coding and agent benchmarks and costs one tenth of GLM-5.2's price.
50%UnverifiedClaude Opus 4.8 will continue to be available to individual Copilot users on annual subscription plans; for request-based annual Pro and Pro+ subscribers, Opus 4.8 has a multiplier of 27 (each request consumes 27 premium requests).
50%UnverifiedIn a preliminary human study involving four experts, the idea sets generated by this model were rated as equally promising as those of Claude Opus 4.8, and slightly more numerous in unique ideas.
50%UnverifiedWhen a coding agent performed proof-of-concept experiments on these ideas, the model's success@k was comparable to that of Claude Opus 4.8.
50%UnverifiedOn a contamination-free benchmark, the trained 8B model surpassed Claude Opus 4.8 and GPT-6 Astra on the creative utility metric.
50%UnverifiedOpus 5 and Opus 4.8 Fast Mode are still in research preview, and their requests still require the beta header.
50%UnverifiedTev-27B achieves Claude Opus 4.8-level performance on the medium-horizon task TB 2.1.
50%UnverifiedThe visual quality scores of AesCode-8B and AesCode-32B are comparable to closed-source frontier models such as GPT-5.5 and Claude Opus 4.8.
50%UnverifiedWithout having seen Terminal-Bench training data, Tev-27B surpassed GLM 5.2 and Gemini 3.7 Flash on the TB 3.0 and 4.0 subsets, and reached the performance of Claude Opus 4.8 on TB 2.1.
50%UnverifiedGLM-5.3-Flash 在 Terminal-Bench 2.1 得84.3分，略低于 Claude Opus 4.8 的85.0分和 GPT-5.6 Terra 的87.4分。
50%UnverifiedGLM-5.3-Flash 在 AutomationBench 上得48.8分，领先 Claude Opus 4.8 的41.0分和 GPT-5.6 Terra 的37.2分，单任务成本约0.045美元。
50%UnverifiedGLM-5.3-Flash 在 Toolathlon Verified 得78.4分，OfficeQA Pro 得62.4分，均领先 Claude Opus 4.8。
50%UnverifiedWithout thinking, even the strongest Claude Opus 4.8 answered nearly half of the numerical questions incorrectly.
50%UnverifiedOn UC-Bench, Qwen3.5-4B trained on UC-Data outperformed larger general-purpose models such as Claude Opus 4.8.
50%UnverifiedParticipants in the experiment included a locally run 4-bit quantized Gemma 26B (built on MLX), Claude Opus 4.8 accessed via API, and GPT-5.4.
50%UnverifiedArtificial Analysis gave Qwen 3.8 an intelligence index of 52, tying GPT at the full reasoning tier at 5.6, with an agent score of 51, higher than Claude Opus 4.8.
50%UnverifiedIn multimodal Agent evaluations, the capabilities of DeepSeek-V-Flash-Vision-XP approach those of Claude Opus 4.8.
50%UnverifiedWorldClaw uses an AI Agent pipeline with Claude Opus 4.8 as the orchestration brain, and calls models such as GPT Image 2, SAM3D, and Hunyuan3D to handle image generation, segmentation, and 3D reconstruction.
50%