Weekly AI Limit Report

01 / THIS WEEK

1. Jul 26

   DeepSeek ↓ DeepSeek V4 Flash input price $0.0938 → $0.14 per 1M tokens (+49%) source

2. Jul 26

   DeepSeek ↓ DeepSeek V4 Flash output price $0.19 → $0.28 per 1M tokens (+49%) source

3. Jul 25

   Cursor → Free tier now advertises Composer access in place of limited Tab completions. The pricing page does not say whether free Tab completions changed, so this is recorded as a swap in what is advertised, not a confirmed gain or loss. · unannounced source

4. Jul 24

   Character.AI → Adds world-building context for Characters; beta with staged early access. source

5. Jul 24

   Claude ↑ New flagship Opus model at identical per-token pricing; Opus 4.8 still listed, so no forced migration or cost increase yet. source

6. Jul 24

   GitHub Copilot ↑ Adds a new model option for users, increasing capability. source

7. Jul 24

   Midjourney ↑ New model version released; default-model status not stated in the announcement. source

8. Jul 24

   Notion AI → Introduces visibility into usage metrics for Notion Workers, preparing users for future billing transitions post-beta. source

9. Jul 23

   Claude ↑ Enterprise tiers get a 500k context window on the default model (Team remains 200k), allowing larger document processing and longer session history. source

10. Jul 23

    GitHub Copilot ↑ Adds autonomous background agent capabilities for Linear issue management. source

11. Jul 23

    GitHub Copilot → Requires migration to stateless MCP protocol for GitHub MCP Server integrations. source

12. Jul 23

    GitHub Copilot ↑ Expands Copilot coding agent capabilities to mobile Actions troubleshooting. source

13. Jul 23

    GitHub Copilot ↑ Introduces review controls for autonomous agent issue management. source

14. Jul 23

    Grok ↑ Adds granular control over audio processing, reducing potential for missed speech in noisy environments. source

15. Jul 23

    v0 → Adds a model to AI Gateway with a time-limited free window that ends mid-August 2026. source

16. Jul 22

    Bolt.new → Adds a new feature for reusable context and workflow management, reducing setup friction. source

17. Jul 22

    Cursor → Addition of a new routing feature, likely impacting model selection or request handling. source

18. Jul 22

    DeepSeek ↓ DeepSeek V4 Flash input price $0.0938 → $0.0980 per 1M tokens (+4%) source

19. Jul 22

    DeepSeek ↓ DeepSeek V4 Flash output price $0.19 → $0.20 per 1M tokens (+4%) source

20. Jul 22

    Lovable → Default-model swap for any app that does not pin a model — output behaviour can shift with no action by the builder (announced in the changelog). source

02 / POST IT

Copy-paste for X, LinkedIn or your newsletter:

What changed in AI tool limits & pricing — past 7 days:

↓ DeepSeek — DeepSeek V4 Flash input price $0.0938 → $0.14 per 1M tokens (+49%) (Jul 26)
↓ DeepSeek — DeepSeek V4 Flash output price $0.19 → $0.28 per 1M tokens (+49%) (Jul 26)
→ Cursor — Free tier now advertises Composer access in place of limited Tab completions. The pricing page does not say whether free Tab completions changed, so this is recorded as a swap in what is advertised, not a confirmed gain or loss. (Jul 25)
→ Character.AI — Adds world-building context for Characters; beta with staged early access. (Jul 24)
↑ Claude — New flagship Opus model at identical per-token pricing; Opus 4.8 still listed, so no forced migration or cost increase yet. (Jul 24)
↑ GitHub Copilot — Adds a new model option for users, increasing capability. (Jul 24)

Dated + sourced tracker → https://ailimit.watch/changes/