Hy3 vs Qwen3.8 Flash
Prices checked
Qwen3.8 Flash takes 3 of 5 rounds against Hy3. Hy3 counters on price: $0.11 per million tokens against $0.19, for 262k context instead of 1M.
TencentHy3
262k context · tool calling · open weights
$0.11/M tokensCheapest price, at NanoGPT · 5.5¢ per 1,000 pages
Alibaba CloudWins 3 roundsQwen3.8 Flash
1M context · tool calling · reads images
$0.19/M tokensCheapest price, at Alibaba Cloud (China) · 9.5¢ per 1,000 pages
Maker's price in (lower wins)
$0.13/M
$0.15/M
Maker's price out (lower wins)
$0.53/M
$0.47/M
Cheapest price (lower wins)
$0.11/M at NanoGPT
$0.19/M at Alibaba Cloud (China)
Maker's price, blended (lower wins)
$0.23/M
$0.23/M
Cost per 1,000 pages (lower wins)
5.5¢
9.5¢
Context
262k tokens
1M tokens
Open weights
yes
no
Tool calling
yes
yes
Structured output
yes
yes
Reads images
no
yes
Reasoning
yes
yes
Released
6 Jul 2026
26 Aug 2026
Providers selling it
29
41
Price we measure against
the maker's own
the maker's own
Trains on your prompts (no is better)
not published
not published
Providers selling both · 17
NanoGPT$0.11 / $0.21
Eden AI$0.23 / $0.23
Impossibl$0.23 / $0.23
Kilo Gateway$0.23 / $0.23
LLM Gateway$0.23 / $0.23
Nous Portal$0.23 / $0.23
OpenRouter$0.23 / $0.23
LLMTR$0.25 / $0.23
Hy3, then Qwen3.8 Flash, per million tokens; the 8 cheapest of 17.
Go further
Add a third or fourth model, or change the weighting, in the tool.
What moved this week
AIHubMix · 2% cheaper · $0.17 / $0.57 $0.16 / $0.62
Frequently asked
- Which is cheaper, Hy3 or Qwen3.8 Flash?
- Hy3, at $0.11 per million tokens against $0.19 for Qwen3.8 Flash, blended 3 in : 1 out, read on 31 Oct 2026.
- Where is each one cheapest?
- Hy3: $0.11 at NanoGPT. Qwen3.8 Flash: $0.19 at Alibaba Cloud (China).
- Can I buy both from one provider?
- Yes. 17 providers sell both: NanoGPT, Eden AI, Impossibl, Kilo Gateway, LLM Gateway, Nous Portal, and more. The cheapest of them for the two together is NanoGPT.
- Which holds more context?
- Qwen3.8 Flash, at 1M tokens against 262k.
Prices exclude tax, read 31 Oct 2026 from each provider's own catalog. They are per M tokens, blended 3 in : 1 out, and a page is 600 tokens in and 60 out. Bars are scaled to the larger value in each row.
Related comparisons