AI Index

Ling-2.6-flash

Language Ant Group Partly verified Released Apr 22, 2026
  • Out Text
  • Reasoning
  • Tool use
  • Open weights

Prices

1 offering · prices as listed
Ant Ling API Official China · CNY Ling-2.6-flash
Conditions/ 1M tokensInputCache readOutput
Standard no longer listed since Oct 20, 2026 ¥0.60¥0.12¥1.80

Prices from Ant Ling: model pricing (checked Oct 19, 2026)

Standard: the price a request with no conditions pays (the headline price elsewhere on the site).

Details

Family
ling-flash
Model ID

Not stated on Ant Group’s pages: max input, input modalities, reasoning efforts, reasoning can be off, max thinking budget, structured output, built-in tools, knowledge cutoff, aliases.

Run locally

Weights: inclusionAI/Ling-2.6-flash
  • Parameters 107.5B
  • Active 7.4B · 256 experts, 8 per token
  • Native precision BF16
  • Layers 32
  • KV cache per token 36 KB
  • Config positions 128K (config.json max_position_embeddings)
  • API context 250K (maker’s API)
My hardware GeForce RTX 4090 24 GB · 1008 GB/s · 64 GB DDR5 dual channel · 89.6 GB/s
On your hardware · GeForce RTX 4090
Needs system memory Q4_K_M (automatic) 8K context 13–20 tok/s23/32 layers in system memory
WeightsKV cacheOverheadIn system memoryUsable
Needs
≈ 64.1 GB (weights 60.8 + KV cache 0.3 + overhead 3.0)
Usable
21.6 GB (90% of its memory)
In system memory
43.7 GB · 9/32 layers on the GPU (--n-gpu-layers 9)
Generation speed
13–20 tok/s (reads the active 7.4B per token)
First token
about 0.24–0.36 s for a 2K prompt
PrecisionWeightsNeeds at 8KOn your hardwareLongest context
BF16 (native)maker’s file200.2 GB≈ 210.5Doesn’t fit—
Q8_0106.4 GB≈ 112.0Doesn’t fit—
FP8maker’s file101.5 GB≈ 106.8Doesn’t fit—
Q6_K82.2 GB≈ 86.6Doesn’t fit—
Q5_K_M71.2 GB≈ 75.0Needs system memory 11–16 tok/s—
Q5_068.8 GB≈ 72.5Needs system memory 11–17 tok/s—
Q4_K_M60.8 GB≈ 64.1Needs system memory 13–20 tok/s—
Q4_056.3 GB≈ 59.4Needs system memory 22–33 tok/s—
INT4 (AWQ / GPTQ)maker’s file60.4 GB≈ 63.7Needs system memory 13–20 tok/s—
IQ4_XS53.2 GB≈ 56.1Needs system memory 23–35 tok/s—
Q3_K_Mquality drops48.9 GB≈ 51.7Needs system memory 25–38 tok/s—
IQ3_XXSquality drops38.3 GB≈ 40.5Needs system memory 33–49 tok/s—
Q2_Kquality drops37.5 GB≈ 39.7Needs system memory 33–50 tok/s—
IQ2_XXSquality drops25.8 GB≈ 27.3Needs system memory 74–111 tok/s—

Click a row to choose a precision. “Maker’s file” sizes are the maker’s own files; others are parameters × bits per weight.

Compared with other hardware

Each at the highest precision it runs. Click one for details.

Smallest that runs it (Q4_K_M, 8K): GPU NVIDIA H100 SXM; unified memory Apple M5 Ultra (64-core GPU) · 96 GB.

Run commands

The maker publishes no files at this precision. Choose its native precision or one of its own builds, or quantise it yourself with llama.cpp.

Memory to fine-tune
Full≈ 1604.3 GBDoesn’t fit your hardware
LoRA≈ 202.9 GBDoesn’t fit your hardware
QLoRA≈ 59.0 GBDoesn’t fit your hardware

Batch 1, 2,048-token sequences, gradient checkpointing; LoRA and QLoRA train rank-16 adapters on the attention projections.

Estimates; facts from the maker’s Hugging Face repositories, as of 2026-11-06. How it’s estimated

Cited scores

From third-party leaderboards, not measured by us

This model isn’t on any leaderboard we cite yet.

Changes

All changes →

No changes recorded yet.

Sources

Some values on this page are still waiting to be checked against official pages: a filled dot is verified, a hollow one is not yet (with or without a source on file). How we verify data

Ant Group · China · developer.ant-ling.com

Snapshots, the same family and other models from this provider