As of August 20, 2026, Hermes 3 70B is 60% cheaper per million tokens than ERNIE 4.5. ERNIE 4.5 is $0.40 / $4.00 per million input/output tokens versus Hermes 3 70B at $0.35 / $1.40. Source: https://anotherwrapper.com/tools/llm-pricing
Source: AnotherWrapper LLM pricing (https://anotherwrapper.com/tools/llm-pricing)
ERNIE 4.5
Baidu
Input
$0.40
Output
$4.00
Hermes 3 70B
Nous Research
Input
$0.35
Output
$1.40
Token pricing, specs, and benchmarks side by side
ERNIE 4.5Hermes 3 70BVerdict
Hermes 3 70B offers both lower pricing and stronger benchmark performance across the board, making it the clear value leader in this comparison.
As of August 20, 2026, Hermes 3 70B is 60% cheaper on blended LLM API pricing. Hermes 3 70B costs $0.35 per million input tokens and $1.40 per million output tokens ($1.75 blended 1M-in + 1M-out). ERNIE 4.5 costs $0.40 / $4.00 per million tokens ($4.40 blended). Sources: https://anotherwrapper.com/tools/llm-pricing and https://anotherwrapper.com/tools/llm-pricing. The cheapest LLM for your app still depends on how many output tokens you generate.
ERNIE 4.5 Baidu API pricing is $0.40 input and $4.00 output per million tokens via Baidu. Hermes 3 70B Nous Research API pricing is $0.35 input and $1.40 output per million tokens via Nous Research. Use those four numbers, not ChatGPT Plus or Claude Pro subscription prices, when you are comparing APIs.
ERNIE 4.5 has a published HumanEval score of 25%. Hermes 3 70B does not have that eval on this page yet.
ERNIE 4.5 supports a 128K token context window, while Hermes 3 70B supports 131K tokens. Hermes 3 70B offers a larger context window. Filling a larger window bills more input tokens, so the cheaper-per-million model can still cost more on long documents.
Hermes 3 70B leads on 10 of 11 shared benchmarks versus ERNIE 4.5's 1 wins. Check the comparison table for GPQA Diamond, SWE-bench, MMLU, HLE, and the other evals we track. Token price and benchmark score together are the usual LLM comparison, not either number alone.
Both ERNIE 4.5 and Hermes 3 70B are production API models. For cost-sensitive production traffic, Hermes 3 70B has better economics on the blended 1M-in + 1M-out scale. For maximum capability, weight the benchmark table for your domain. Many production stacks route cheap models for drafts and a frontier model for the hard turn, which is usually cheaper than sending everything to the expensive API.
Yes. Call ERNIE 4.5 through Baidu and Hermes 3 70B through Nous Research with separate API keys, or through a gateway that already wraps both. Keep prompts in tokens, not characters, when you estimate the invoice. AnotherWrapper templates can swap providers without rewriting auth and billing.
List rates are the latest published Baidu API pricing and Nous Research API pricing, quoted per million tokens. Official rate sheets: https://anotherwrapper.com/tools/llm-pricing and https://anotherwrapper.com/tools/llm-pricing. Prompt caching, batch APIs, and committed-use discounts can change the invoice. Benchmark scores are from official publications and independent evals, updated as new numbers land.
ERNIE 4.5 supports up to 66K output tokens per request, while Hermes 3 70B supports up to 16K output tokens. Output tokens are usually the expensive half of API pricing, so a higher max output cap is a capability, not a discount.
Neither ERNIE 4.5 nor Hermes 3 70B has published throughput data on this page yet. Throughput (tokens per second) and time-to-first-token are why two models with the same per-million-token price can still feel different in chat and agents.
Use the model pickers at the top of this LLM comparison. Search any model in the index and the URL updates. Open ERNIE 4.5 or Hermes 3 70B from the sidebar for the single-model API pricing page, or go back to the LLM pricing table to sort by cheapest blended cost.
This comparison covers API pricing, context window, throughput, and benchmark scores. ERNIE 4.5 costs $0.40/M input and $4.00/M output. Hermes 3 70B costs $0.35/M input and $1.40/M output.
Rank both on the LLM leaderboard, SWE-bench, and Humanity's Last Exam.
The index
280 models across 33 providers. Search or jump to a lab — every model page stays linked here.
Next step
Auth, billing, and the API layer are already decided. Fork 8 finished AI apps, or have us build the first version with you.
