Cheapest LLM APIs, Ranked by Price

Every LLM API on A2Agent, cheapest first by blended price (input and output weighted 3:1).

Struck-through is the vendor’s official price. Bold is what your API key pays.

Model	Platform	Context	Input /1M	Output /1M	Discount
DeepSeek V4 Flash	DeepSeek	1M	~~$0.140~~$0.084	~~$0.280~~$0.168	-40%
GLM-5.3 Flash	Z.ai	1M	~~$0.150~~$0.105	~~$0.500~~$0.350	-30%
Qwen3.8 Flash	Qwen	1M	~~$0.150~~$0.135	~~$0.470~~$0.423	-10%
MiniMax M3.1 Flash	MiniMax	1M	~~$0.150~~$0.135	~~$0.600~~$0.540	-10%
DeepSeek V4.1 Flash	DeepSeek	1M	~~$0.270~~$0.162	~~$0.810~~$0.486	-40%
Qwen3.7 Flash	Qwen	1M	~~$0.200~~$0.140	~~$0.800~~$0.560	-30%
MiniMax M3	MiniMax	1M	~~$0.300~~$0.150	~~$1.20~~$0.600	-50%
Qwen3.6 Flash	Qwen	1M	~~$0.200~~$0.140	~~$1.20~~$0.840	-30%
DeepSeek V4 Pro	DeepSeek	1M	~~$0.435~~$0.261	~~$0.870~~$0.522	-40%
Qwen3.5 Plus	Qwen	1M	~~$0.500~~$0.350	~~$3.00~~$2.10	-30%
Kimi K2.5	Moonshot	256K	~~$0.600~~$0.420	~~$3.00~~$2.10	-30%
GLM-5	Z.ai	200K	~~$1.00~~$0.600	~~$3.20~~$1.92	-40%
GLM-5.1	Z.ai	200K	~~$1.15~~$0.690	~~$4.00~~$2.40	-40%
Kimi K2.7 Code	Moonshot	256K	~~$0.950~~$0.665	~~$4.00~~$2.80	-30%
GLM-5.2	Z.ai	1M	~~$1.40~~$0.840	~~$4.40~~$2.64	-40%
GLM-5.3	Z.ai	1M	~~$1.40~~$0.840	~~$4.40~~$2.64	-40%
Qwen3.7 Plus	Qwen	1M	~~$1.20~~$0.960	~~$4.80~~$3.84	-20%
Qwen3.7 MAX	Qwen	1M	~~$1.70~~$1.36	~~$5.10~~$4.08	-20%
Qwen3.8 MAX	Qwen	1M	~~$2.00~~$1.80	~~$6.00~~$5.40	-10%
Kimi K3	Moonshot	1M	~~$3.00~~$2.40	~~$15.00~~$12.00	-20%

USD per 1M tokens. Final billing depends on your API key group and metered usage.

Where to start in each family

DeepSeek API

Start on Flash, the cheap line. Pro costs more, and a spec sheet won’t tell you whether it’s worth it. Run your own prompts and move up when Flash keeps getting them wrong.

GLM API (Z.ai)

Flash is the bargain. It reads images and has the long window, at a price far under the numbered models. Those carry the coding and agent tags, and only the newer ones get the long window.

Kimi API (Moonshot)

The K2 models share one window size. Plain K2 reads images and K2 Code is tagged for coding. K3 is priced well above the rest. Pay for it when you need the longer window and not before.

MiniMax API

Every model here costs the same, so price settles nothing. Check the context window first, then pick between the coding tag and the agent tag.

Qwen API

Take the newest Flash. Each release has come in cheaper than the last, and all of them read images. MAX carries the reasoning tag and the highest rates. Plus is priced between the two and reads images too.