The 30-second summary
+ What we liked
- Claude Code focus
- New station with competitive entry pricing
− What we didn't
- Very limited public data and reviews
- Unproven reliability
In-depth review
Overview
哈基米 (Hajimi) is a new entrant in the AI token relay space, positioning itself aggressively with a focus on Claude Code and early access to GPT-5.x. As of early 2026, this is a high-risk, high-reward option for developers who want to be on the bleeding edge without VPN access. With a safety rating of 3/5 and 94% uptime, it’s clearly a beta-stage service. If you’re comfortable with occasional instability and want to test unreleased models, this might be your playground. If you need production reliability, look elsewhere.
Pricing & Free Trial
Hajimi currently offers a free tier with no monthly cost, which is unusual for a relay station. This suggests they’re still building their user base and refining infrastructure.
| Plan | Price | Notes |
|---|---|---|
| Free | $0/month | Includes access to Claude Code and GPT-5.x (usage limits likely apply) |
| Pro (speculated) | Not disclosed | Expected to launch with higher rate limits and priority support |
The free tier is attractive for prototyping, but expect throttling. There is no public pricing for paid plans yet—likely coming as they stabilize.
Models & API Compatibility
Hajimi’s standouts:
- Claude Code: Full access via relay, no VPN needed. This is a rare find—most relay stations focus on GPT or Gemini. If you need Claude Code for agentic coding tasks, Hajimi is one of the few options.
- GPT-5.x: Early access to the latest GPT iteration. Note: this is likely a preview, not the stable release. Expect breaking changes and lower reliability.
API compatibility is standard OpenAI/Anthropic-style endpoints. You’ll need to adjust base URLs and API keys. No custom SDKs—just raw HTTP calls.
Max tokens: 0 — this is suspicious. Either it means no limit (unlikely) or the data is missing. In practice, expect context windows similar to the underlying models (e.g., 128K for Claude Code, 128K for GPT-5.x).
Uptime & Reliability
94% uptime is below industry standard (most relay stations promise 99%+). This means roughly 1.5 days of downtime per month. For a development tool, that’s acceptable for testing but not for customer-facing apps.
- Downtime patterns: Likely during model upgrades or infrastructure scaling. No public status page yet.
- Latency: Unknown, but with 94% uptime, expect occasional timeouts or 503s.
China Access
Works without VPN. Hajimi explicitly targets Chinese developers, so expect CDN nodes in Asia. No reports of IP blocking or throttling from mainland China, but with limited user data, this is unverified.
Security & Safety
Safety rating 3/5 — average. They likely implement basic rate limiting and API key authentication but lack advanced features like IP whitelisting or audit logs. Do not use for production workloads until they improve.
Pros & Cons
Pros
- Claude Code access: Rare among relay stations. Essential for agentic coding.
- GPT-5.x preview: Early adopter advantage.
- Free tier: Zero cost to test.
- No VPN needed: Works from mainland China.
Cons
- Low uptime (94%): Unreliable for production.
- Limited reviews: Almost no community feedback.
- No paid plan details: Pricing unclear for scaling.
- Safety rating 3/5: Basic security only.
Verdict
Hajimi is a bleeding-edge tester’s tool, not a production service. If you’re a developer in China who needs to experiment with Claude Code or GPT-5.x without a VPN, the free tier is worth a spin. But don’t bet your workflow on it—expect outages, no support, and potential model instability. Monitor their uptime and wait for paid tiers and community reviews before committing.
Recommendation: Use for prototyping only. Have a backup relay station (e.g., API2D or GPT_API_free) for when Hajimi goes down.
FAQ
Q: Can I use Hajimi for production apps?
A: No. With 94% uptime and no paid plan details, it’s not production-ready. Use for testing or personal projects only.
Q: Does Hajimi support streaming responses?
A: Likely yes, given standard API compatibility, but not confirmed. Test with the free tier.
Q: How do I get GPT-5.x access?
A: Sign up for the free tier. GPT-5.x is listed as available, but expect it to be a preview with potential rate limits.
Q: Is my API key safe with Hajimi?
A: Safety rating 3/5 suggests basic encryption but no advanced security. Use a dedicated API key with limited permissions.
Q: What happens if Hajimi shuts down?
A: High risk for a new station. Keep backups of your prompts and switch to another relay if needed. No migration tools available.
Pricing breakdown
哈基米 offers competitive pricing for developers. Here's the breakdown:
| Plan | Price | Quota | Best for |
|---|---|---|---|
| Free | $0/mo | Free trial | Kicking the tires |
| Standard RECOMMENDED | Pay-as-you-go/mo | Unlimited usage | Solo devs · small teams |
| Enterprise | Custom | SLA · dedicated support | Teams & agencies |
Supported models
2 models across major vendors.
Frequently asked questions
Can I access this platform from China without a VPN?
Most relay stations are accessible from Chinese ISPs. Check our review for specific routing details.
What payment methods are accepted?
Payment options vary by platform. Some accept Alipay/WeChat Pay, others are USD/crypto only.
How does this compare to using OpenAI directly?
Relay stations add routing latency but provide access from restricted regions, unified billing, and multi-model fallback.
Is my API key safe?
Keys are encrypted at rest. Most platforms support per-project scoping and IP allow-lists.
Should you use 哈基米?
Bleeding-edge testers only