OPEN LINK ↗
// 73d agoPRODUCT UPDATE
Claude Opus 4.6, Sonnet 4.6 unlock 1M context, flat pricing
Anthropic makes 1M-token context generally available for Opus 4.6 and Sonnet 4.6 with no long-context pricing surcharge — eliminating the previous 2x cost multiplier above 200K tokens. Opus 4.6 scores 78.3% on MRCR v2, the highest frontier long-context retrieval result at 1M tokens recorded.
// ANALYSIS
Killing the long-context premium is the real story here — it turns 1M context from a niche expensive capability into a default pricing tier that enterprises can actually budget around.
- –Flat pricing ($5/$25 for Opus 4.6, $3/$15 for Sonnet 4.6) per million tokens regardless of context length undercuts both Gemini 3.1 Pro and GPT-5.4, which both charge more beyond their base context thresholds
- –MRCR v2 score of 78.3% at 1M tokens is a fourfold jump over the prior generation's 18.5% at that length — retrieval quality at scale actually holds up
- –Media capacity jumps from 100 to 600 images or PDF pages per request, unlocking document-heavy pipelines
- –Claude Code's Max/Team/Enterprise tiers get Opus 4.6 at 1M context automatically, with no API flag required
- –Adaptive reasoning controls (four granular effort levels) and 128K max output for Opus 4.6 make the model significantly more viable for long agentic sessions
// TAGS
claudellminferenceapiagentbenchmark
DISCOVERED
73d ago
2026-03-15
PUBLISHED
73d ago
2026-03-15
RELEVANCE
9/ 10
AUTHOR
DIY Smart Code